CN118505516A

CN118505516A - Method and system for enhancing consultation technology through panoramic interaction display

Info

Publication number: CN118505516A
Application number: CN202410958381.8A
Authority: CN
Inventors: 陈赟; 屈定春; 谢志伟; 张伟; 戴凯; 焦宏; 王春兵
Original assignee: Gongcheng Management Consulting Co ltd
Current assignee: Gongcheng Management Consulting Co ltd
Priority date: 2024-07-17
Filing date: 2024-07-17
Publication date: 2024-08-16
Anticipated expiration: 2044-07-17
Also published as: CN118505516B

Abstract

The application relates to the technical field of computer image processing, in particular to a method and a system for enhancing consultation technology through panoramic interaction display, through a series of steps of acquiring multimode image descriptors, determining original enhanced display tracking objects, calculating enhanced rendering confidence weights, determining optimization instructions and the like, not only are the influence of the objects in the image quantized, but also scientific basis is provided for subsequent image optimization. Finally, according to the panoramic monitoring image optimization indication determined by the weights, the visual effect and the information transmission capability of the panoramic image can be remarkably improved, so that engineering management personnel can acquire key information more intuitively and rapidly, and further accurate judgment and decision can be made. Therefore, the recognition precision, tracking efficiency and information presentation quality of the engineering management panoramic monitoring image can be improved, and the efficiency and the intelligent degree of engineering management are improved.

Description

Method and system for enhancing consultation technology through panoramic interaction display

Technical Field

The application relates to the technical field of computer image processing, in particular to a method and a system for enhancing consultation technology through panoramic interaction display.

Background

In the field of engineering management, panoramic monitoring images are increasingly widely applied, can provide comprehensive view angles and rich field information, and are important for ensuring engineering quality and safety. However, the conventional panoramic monitoring image processing method has certain limitations in terms of identification precision, tracking efficiency, information presentation and the like, and cannot meet the requirements of modern engineering management on high precision and high efficiency.

For example, conventional image processing methods often fail to accurately capture the various interactive elements of an image, resulting in inaccurate identification and tracking of key elements. Meanwhile, due to lack of scientific quantification basis, the optimization processing of the image is often performed empirically, and the best effect is difficult to achieve. These problems not only affect the accuracy and efficiency of engineering management, but may also present potential safety hazards.

Disclosure of Invention

In order to improve the problems, the application provides a method and a system for enhancing consultation technology through panoramic interactive display.

The embodiment of the application provides a method for enhancing consultation technology by panoramic interaction display, which is applied to a panoramic interaction processing system and comprises the following steps: acquiring a multimode image descriptor of an initial engineering management panoramic monitoring image, wherein the multimode image descriptor of the initial engineering management panoramic monitoring image is used for representing an image interaction element vector of the initial engineering management panoramic monitoring image; determining at least one original enhanced presentation tracking object in the initial engineering management panoramic monitoring image based on a multimode image descriptor of the initial engineering management panoramic monitoring image; determining enhanced rendering confidence weights of the original enhanced display tracking objects, wherein the enhanced rendering confidence weights are used for representing influence coefficients of the original enhanced display tracking objects in the initial engineering management panoramic monitoring image; and determining panoramic monitoring image optimization instructions of the initial engineering management panoramic monitoring images according to the enhanced rendering confidence weights of the original enhanced display tracking objects.

Under some technical ideas, the determining the enhanced rendering confidence weight of each original enhanced presentation tracking object includes: acquiring pixel cluster wavelet characteristics of each original enhanced display tracking object; and determining the enhanced rendering confidence weight of each original enhanced display tracking object according to the pixel cluster wavelet characteristics of each original enhanced display tracking object and the multimode image descriptors of the initial engineering management panoramic monitoring image.

Under some technical ideas, the multimode image descriptor of the initial engineering management panoramic monitoring image comprises first linear embedded knowledge, wherein the first linear embedded knowledge is used for representing an image interaction element vector of the initial engineering management panoramic monitoring image; the determining the enhanced rendering confidence weight of each original enhanced display tracking object according to the pixel cluster wavelet characteristics of each original enhanced display tracking object and the multimode image descriptors of the initial engineering management panoramic monitoring image comprises the following steps: and determining the enhanced rendering confidence weight of each original enhanced display tracking object according to the commonality evaluation between the pixel cluster wavelet characteristics of each original enhanced display tracking object and the first linear embedded knowledge.

Under some technical ideas, determining a panoramic monitoring image optimization indication of the initial engineering management panoramic monitoring image according to the enhanced rendering confidence weights of the original enhanced display tracking objects comprises: determining an original enhanced display tracking object with the enhanced rendering confidence weight meeting a first judging requirement as an enhanced display tracking object contained in the initial engineering management panoramic monitoring image; wherein the first determination requirement includes a minimum one of: the enhanced rendering confidence weight is not smaller than a first preset value; and sorting all the original enhanced display tracking objects according to the descending order of the enhanced rendering confidence weights, wherein p bits are positioned before queuing, and p is a positive integer.

Under some technical ideas, determining a panoramic monitoring image optimization indication of the initial engineering management panoramic monitoring image according to the enhanced rendering confidence weights of the original enhanced display tracking objects comprises: determining the original enhanced display tracking object with the enhanced rendering confidence weight meeting the second judging requirement as a key enhanced display tracking object contained in the initial engineering management panoramic monitoring image; wherein the second determination requirement includes a minimum one of: the enhanced rendering confidence weight is not smaller than a second preset value; and sorting all the original enhanced display tracking objects according to the descending order of the enhanced rendering confidence weights, wherein q bits are positioned before queuing, and q is a positive integer.

Under some technical ideas, the determining at least one original enhanced presentation tracking object in the initial engineering management panoramic monitoring image based on the multimode image descriptor of the initial engineering management panoramic monitoring image comprises: determining monitoring event label information of the initial engineering management panoramic monitoring image based on a multimode image descriptor of the initial engineering management panoramic monitoring image, wherein the monitoring event label information of the initial engineering management panoramic monitoring image is used for representing labels of monitoring events of the initial engineering management panoramic monitoring image; and determining at least one original enhanced display tracking object in the initial engineering management panoramic monitoring image based on the multimode image descriptor of the initial engineering management panoramic monitoring image and monitoring event label information of the initial engineering management panoramic monitoring image.

Under some technical ideas, the determining monitoring event tag information of the initial engineering management panoramic monitoring image based on the multimode image descriptor of the initial engineering management panoramic monitoring image includes: determining discrimination possibility corresponding to each of X monitoring event labels based on the multimode image descriptors of the initial engineering management panoramic monitoring image; the judging possibility corresponding to a u-th monitoring event tag in the X monitoring event tags is used for indicating the confidence that the initial engineering management panoramic monitoring image belongs to the u-th monitoring event tag, X is an integer greater than 1, and u is a positive integer not greater than X; based on the discrimination possibility respectively corresponding to the X monitoring event labels, v monitoring event labels with the maximum discrimination possibility are selected from the X monitoring event labels, v target monitoring event labels are obtained, and v is a positive integer; based on the discrimination possibility corresponding to the v target monitoring event labels, splicing the mapping characteristics corresponding to the v target monitoring event labels respectively to obtain mapping splicing characteristics; the monitoring event label information of the initial engineering management panoramic monitoring image comprises the mapping splicing characteristic.

Under some technical ideas, the multimode image descriptor of the initial engineering management panoramic monitoring image comprises linear embedded knowledge corresponding to a plurality of image blocks in the initial engineering management panoramic monitoring image respectively; the determining at least one original enhanced display tracking object in the initial engineering management panoramic monitoring image based on the multimode image descriptor of the initial engineering management panoramic monitoring image and the monitoring event label information of the initial engineering management panoramic monitoring image comprises the following steps: combining the linear embedded knowledge corresponding to each image block in the initial engineering management panoramic monitoring image with monitoring event label information of the initial engineering management panoramic monitoring image to obtain linear embedded combined knowledge corresponding to each image block; based on the linear embedded combined knowledge, at least one original enhanced presentation tracking object in the initial engineering management panoramic monitoring image is determined.

Under some technical ideas, the panoramic monitoring image optimization indication of the initial engineering management panoramic monitoring image is determined by a depth residual memory network, wherein the depth residual memory network comprises an image description mining branch, a tracking object positioning branch and an enhanced rendering processing branch; the image description mining branch is used for acquiring multimode image descriptors of the initial engineering management panoramic monitoring image; the tracking object positioning branch is used for determining at least one original enhanced display tracking object in the initial engineering management panoramic monitoring image based on a multimode image descriptor of the initial engineering management panoramic monitoring image; the enhanced rendering processing branch is used for determining enhanced rendering confidence weights of the original enhanced display tracking objects.

Under some technical ideas, the debugging method of the depth residual memory network comprises the following steps: acquiring a multimode image descriptor of an engineering management panoramic monitoring image example through the image description mining branch, wherein the multimode image descriptor of the engineering management panoramic monitoring image example is used for representing an image interaction element vector of the engineering management panoramic monitoring image example; determining, by the tracking object positioning branch, an enhanced presentation tracking object recognition result of the engineering management panoramic monitoring image example based on a multimode image descriptor of the engineering management panoramic monitoring image example, the enhanced presentation tracking object recognition result including at least one original enhanced presentation tracking object in the engineering management panoramic monitoring image example; determining enhanced rendering confidence weights of the original enhanced display tracking objects through the enhanced rendering processing branches, wherein the enhanced rendering confidence weights are used for representing influence coefficients of the original enhanced display tracking objects in the engineering management panoramic monitoring image example; and determining a key tracking object identification result of the engineering management panoramic monitoring image example according to the enhanced rendering confidence weights of the original enhanced display tracking objects, wherein the key tracking object identification result comprises the following steps: at least one key enhanced presentation tracking object determined from the at least one original enhanced presentation tracking object based on the enhanced rendering confidence weight; and debugging the depth residual memory network based on the enhanced display tracking object recognition result, the key tracking object recognition result, the enhanced display tracking object priori basis and the key tracking object priori basis of the engineering management panoramic monitoring image example.

Under some technical ideas, the debugging the depth residual memory network based on the enhanced display tracking object recognition result, the key tracking object recognition result, the enhanced display tracking object priori basis and the key tracking object priori basis of the engineering management panoramic monitoring image example comprises the following steps: determining a first training error based on the enhanced display tracking object recognition result and the enhanced display tracking object prior basis, wherein the first training error is used for evaluating the quality scores of the tracking object positioning branches; determining a second training error based on the key tracking object recognition result and the key tracking object prior basis, wherein the second training error is used for evaluating the quality score of the enhanced rendering processing branch; and debugging the depth residual memory network based on the first training error and the second training error.

Under some independent technical ideas, the enhanced display tracking object recognition result comprises a likelihood training recognition result corresponding to each image block in the engineering management panoramic monitoring image example, wherein the likelihood training recognition result comprises training prediction confidence degrees of the image blocks relative to a plurality of object types; the enhanced display tracking object prior basis comprises likelihood prior knowledge corresponding to each image block in the engineering management panoramic monitoring image example, wherein the likelihood prior knowledge comprises prior confidence degrees of the image blocks relative to a plurality of object types; the determining a first training error based on the enhanced presentation tracking object recognition result and the enhanced presentation tracking object prior basis includes: and determining the first training error based on the likelihood training recognition results of the respective corresponding image blocks in the engineering management panoramic monitoring image example and the likelihood priori knowledge of the respective corresponding image blocks in the engineering management panoramic monitoring image example.

Under some independent technical ideas, the key tracking object prior basis comprises at least one prior key enhancement display tracking object in the engineering management panoramic monitoring image example; the determining a second training error based on the key tracking object recognition result and the key tracking object prior basis includes: and determining the second training error based on the enhanced rendering confidence weights respectively corresponding to the original enhanced display tracking objects in the engineering management panoramic monitoring image example and the priori critical enhanced display tracking objects in the original enhanced display tracking objects.

Under some independent technical ideas, the depth residual memorization network further comprises a monitoring event decision branch, and the method further comprises: determining a monitoring event tag identification result of the engineering management panoramic monitoring image example based on a multimode image descriptor of the engineering management panoramic monitoring image example through the monitoring event decision branch, wherein the monitoring event tag identification result of the engineering management panoramic monitoring image example comprises discrimination possibility prediction results corresponding to X monitoring event tags respectively; the judging possibility prediction result corresponding to a u-th monitoring event tag in the X monitoring event tags is used for indicating an initial probability value of the engineering management panoramic monitoring image example belonging to the u-th monitoring event tag, X is an integer greater than 1, and u is a positive integer not greater than X; determining monitoring event label information of the engineering management panoramic monitoring image example based on a monitoring event label identification result of the engineering management panoramic monitoring image example, wherein the monitoring event label information of the engineering management panoramic monitoring image example is used for representing a label to which a monitoring event of the engineering management panoramic monitoring image example belongs; the tracking object positioning branch is used for determining an enhanced display tracking object recognition result of the engineering management panoramic monitoring image example based on the multimode image descriptor of the engineering management panoramic monitoring image example and monitoring event label information of the engineering management panoramic monitoring image example; determining a third training error based on the monitoring event tag recognition result and the monitoring event tag priori basis, wherein the third training error is used for evaluating the quality scores of the monitoring event decision branches; the debugging of the depth residual memory network based on the first training error and the second training error comprises: and debugging the depth residual memory network based on the first training error, the second training error and the third training error.

Under some independent technical ideas, the prior basis of the monitoring event labels comprises prior discrimination possibilities corresponding to the X monitoring event labels respectively, and the prior discrimination possibility corresponding to the u-th monitoring event label is used for representing the prior probability value of the engineering management panoramic monitoring image example belonging to the u-th monitoring event label; the determining a third training error based on the monitoring event tag identification result and the monitoring event tag prior basis includes: and determining the third training error based on the discrimination possibility prediction results respectively corresponding to the X monitoring event labels and the priori discrimination possibility respectively corresponding to the X monitoring event labels.

The embodiment of the application provides a panoramic interaction processing system, which comprises at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored by the memory, causing the at least one processor to perform the method described above.

Embodiments of the present application provide a computer readable storage medium having stored thereon a computer program which, when run, implements the method described above.

The embodiment of the application not only quantifies the influence of the objects in the image, but also provides scientific basis for the subsequent image optimization through a series of steps of acquiring the multimode image descriptor, determining the original enhanced display tracking object, calculating the enhanced rendering confidence weight, determining the optimization instruction and the like. Finally, according to the panoramic monitoring image optimization indication determined by the weights, the visual effect and the information transmission capability of the panoramic image can be remarkably improved, so that engineering management personnel can acquire key information more intuitively and rapidly, and further accurate judgment and decision can be made. Therefore, the recognition precision, tracking efficiency and information presentation quality of the engineering management panoramic monitoring image can be improved, and the efficiency and the intelligent degree of engineering management are improved.

Drawings

FIG. 1 is a flowchart of a method for enhancing consultation techniques for panoramic interactive presentation according to an embodiment of the present application;

fig. 2 is a schematic structural diagram of a panoramic interaction processing system 200 according to an embodiment of the present application.

Detailed Description

In order to better understand the above technical solutions, the following detailed description of the technical solutions of the present application is made by using the accompanying drawings and specific embodiments, and it should be understood that the specific features of the embodiments and the embodiments of the present application are detailed descriptions of the technical solutions of the present application, and not limiting the technical solutions of the present application, and the technical features of the embodiments and the embodiments of the present application may be combined with each other without conflict.

Fig. 1 shows a method of panoramic interactive presentation enhanced consultation technology applied to a panoramic interactive processing system, the method including the following steps 110-140.

Step 110, acquiring a multimode image descriptor of an initial engineering management panoramic monitoring image, wherein the multimode image descriptor of the initial engineering management panoramic monitoring image is used for representing an image interaction element vector of the initial engineering management panoramic monitoring image;

Step 120, determining at least one original enhanced display tracking object in the initial engineering management panoramic monitoring image based on the multimode image descriptor of the initial engineering management panoramic monitoring image;

130, determining enhanced rendering confidence weights of the original enhanced display tracking objects, wherein the enhanced rendering confidence weights are used for representing influence coefficients of the original enhanced display tracking objects in the initial engineering management panoramic monitoring image;

And 140, determining panoramic monitoring image optimization instructions of the initial engineering management panoramic monitoring images according to the enhanced rendering confidence weights of the original enhanced display tracking objects.

For easy understanding, how the panoramic interactive processing system performs the above technical solution (the technical solution may also be understood as a panoramic interactive presentation enhancement image processing method based on artificial intelligence) will be described in detail below by way of an application scenario example.

In a large building engineering project, a panoramic interactive processing system is introduced to improve the efficiency and accuracy of engineering management. The panoramic interaction processing system first performs step 110 to obtain a multimode image descriptor of the initial engineering management panoramic monitoring image. The multimode image descriptor is extracted from the panoramic monitoring image through advanced image processing technology, and can fully represent the interaction element vectors in the image, such as key characteristics of color, texture, shape and the like.

Next, in step 120, the panoramic interaction processing system uses the multimode image descriptors to intelligently identify at least one original enhanced presentation tracking object in the initial engineering management panoramic monitoring image. The original enhanced presentation tracking object may include key engineering management elements of the building structure being constructed, important equipment, worker activities, etc.

Proceeding to step 130, the panoramic interaction processing system proceeds to conduct an in-depth analysis of each identified original enhanced presentation tracking object to determine their enhanced rendering confidence weights. The enhanced rendering confidence weight is comprehensively calculated according to a plurality of factors such as importance and significance of the object in the panoramic image, correlations with other objects and the like. For example, a critical structural portion being constructed may be given a higher weight because it has a critical impact on the overall progress and safety of the project.

Finally, in step 140, the panoramic interaction processing system generates a panoramic monitoring image optimization designation based on the enhanced rendering confidence weights for each of the original enhanced presentation tracking objects. Panoramic monitoring image optimization directives may include applying different visual effect enhancement techniques, such as highlighting, dynamic labeling, or adding interactive information, etc., to different objects so that project managers can more intuitively understand project progress and potential problems.

By the method, the panoramic interaction processing system not only provides immersive panoramic browsing experience, but also provides powerful decision support for engineering management staff through intelligent analysis and panoramic image optimization, so that the efficiency and accuracy of engineering management are remarkably improved, and meanwhile, communication and cooperation among project teams are promoted.

In addition, in practical implementation, the panoramic interaction processing system can be integrated with project management software to realize data sharing and real-time updating. Engineering management personnel can view the latest panoramic monitoring image and the optimizing instruction at any time through the system, so that project progress and quality control points can be better mastered. In addition, the panoramic interaction processing system also supports multi-equipment adaptation cooperative work, so that team members in different places can access and update project information at any time and any place, and convenience and flexibility of engineering management are further improved.

In connection with the above application scenario, the following description will be made with respect to steps 110 to 140, respectively.

In step 110, the initial engineering panoramic monitoring image refers to a first panoramic image captured by panoramic photography at the beginning of an engineering project. The image not only covers the entire scene of the engineering project, but also includes all relevant details such as building structure, equipment layout, construction environment, etc. This image will be used as a benchmark for engineering management for subsequent monitoring and comparison. For example, in a large commercial complex project, the initial project management panoramic monitoring image may include a panoramic view of the entire worksite, clearly visible from the excavation of the pit to the rebar structure that has been erected.

The multimode image descriptor refers to image feature descriptions of multiple modes extracted from the panoramic monitoring image. The descriptors can capture various information such as color, texture, shape and the like of the image and convert the information into mathematical expression forms which can be understood and compared by a computer. The multimode image descriptor includes not only conventional image features (such as SIFT, SURF, etc.), but also high-level features extracted by deep learning techniques. For example, in a job site image containing multiple building materials, the multimode image descriptors can accurately identify and describe the texture features of different materials, such as rough surfaces of concrete, smooth texture of steel materials, and the like.

The image interaction element vector refers to a form of converting key elements (such as building structures, equipment, personnel and the like) in the panoramic monitoring image into numerical feature vectors. These vectors may be combined from features extracted from the multimode image descriptors and may be advantageously used for processing and analysis in computer vision algorithms. For example, an image interaction element vector may include edge contour information, color distribution, texture features, etc. of a building structure, all quantized to specific values, forming a multi-dimensional vector. Such vectors can be used not only for recognition and tracking of objects, but also for analyzing relationships and dynamic changes between objects.

In detail, based on step 110, the panoramic interaction processing system first loads an initial engineering management panoramic monitoring image, which is a high definition panoramic image containing the entire engineering project scene. The system then uses advanced image processing techniques, such as deep learning algorithms and feature extraction methods, to carefully analyze the panoramic view. The system generates a multimodal image descriptor by extracting various features in the image, such as color, texture, shape, etc. These descriptors can reflect the various elements and their features in the image comprehensively and accurately. For example, the system may identify different building materials in the image and generate corresponding descriptors based on their color, texture, etc. Further, these multimode image descriptors are converted into image interaction element vectors. This is a process by which the image features are quantized, whereby each element in the image is assigned a specific numerical feature vector. These vectors include not only the characteristic information of the elements themselves, but also the spatial relationships and interactions between the elements. Finally, the panoramic interaction processing system stores these image interaction element vectors for use in subsequent steps. These vectors will become the key basis for the system to intelligently identify, track and optimize the engineering management elements in the panoramic image. By the method, the system can grasp the actual condition of the engineering project more accurately, and more accurate and comprehensive information support is provided for engineering management staff.

In step 120, the original enhanced presentation tracking object refers to a key target identified and selected from the initial engineering management panoramic monitoring image. These objects may be elements that require special attention and tracking during engineering management, such as building structures under construction, critical equipment, personnel activity areas, etc. After the objects are identified in the panoramic image, the objects are subjected to enhancement processing of a system, such as means of adding visual identification, dynamic effects or interactive information, so that the objects are more prominent in panoramic display, and engineering management personnel can conveniently and quickly locate and know the real-time states and changes of the key objects. For example, a high-rise building frame under construction can be identified and set as an original enhanced display tracking object, and the original enhanced display tracking object is highlighted or labeled in the panoramic image through the processing of the system, so that engineering management personnel can monitor and track the original enhanced display tracking object continuously.

Upon execution to step 120, the system begins to identify and determine the original enhanced presentation tracking object in the image using its previously extracted multimode image descriptors of the initial engineering management panoramic monitoring image. The descriptors contain key characteristic information such as color, texture, shape and the like of the image, and are the basis of intelligent recognition of important objects in the image by the system. The system firstly carries out deep analysis and processing on the descriptors, and identifies key engineering management elements which can exist in the panoramic image by comparing and matching a preset engineering management key object feature library. These elements may include the building structure being constructed, important equipment, personnel-intensive areas, etc., which are the objects of major attention and tracking in the engineering management process. During the recognition process, the system will also score and rank each identified object according to its specific features and importance to determine which objects are the most needed to be enhanced for presentation and tracking. For example, a critical load bearing structure under construction may be given a higher score due to its importance and thus be preferentially selected as the original enhanced display tracking object. Once the system has determined these objects, it proceeds to the next process, enhanced presentation and continuous tracking of these objects. This includes adding visual identification, dynamic effects, or interactive information, etc. to these objects in the panoramic image so that engineering administrators can more intuitively understand and monitor the real-time status and changes of these key objects. By the method, the panoramic interaction processing system greatly improves the efficiency and accuracy of engineering management, and provides powerful technical support for smooth implementation of engineering projects.

In step 130, the enhanced rendering confidence weight is an important parameter that represents the importance and significance of the original enhanced presentation tracking object in the panoramic monitoring image. This weight is calculated based on a combination of factors including, but not limited to, the location of the object in the image, size, color contrast, and degree of association with the surrounding environment. A high weight means that the object has a higher visual appeal and information content in the panoramic image, and therefore should be given more attention and prominence when performing image optimization and enhanced presentation. For example, a large building structure occupying the center of an image may be given a higher enhanced rendering confidence weight due to its significant visual features and significant engineering management significance.

The influence coefficient is an index for quantifying the influence degree of the original enhanced display tracking object on the whole visual effect and engineering management meaning in the panoramic monitoring image. It is closely related to the enhanced rendering confidence weight, which can be said to be the basis for determining the influence coefficient. The high influence coefficient of an object means that the representation of the object in the panoramic image influences the understanding and judgment of the scene by engineering management staff to a greater extent. Thus, in performing the rendering and optimization of the panoramic image, the system adjusts its visual effect according to the influence coefficient of each object to ensure that important engineering management elements can get sufficient attention and prominence.

The panoramic interaction processing system, when executing step 130, will integrate a number of factors to determine the enhanced rendering confidence weights for each original enhanced presentation tracking object. This step is a key element in the image optimization process, and is directly related to the final presentation effect of the panoramic image and the accuracy of information obtained by engineering management personnel. The system first analyzes the location of each object in the panoramic image. An object located in the center or significant position of the image may be more noticeable and therefore may be given a higher weight. Next, the system evaluates the size and shape of the object, and large or uniquely shaped objects tend to have a greater impact on the overall visual effect, so that higher weights can also be obtained. In addition, color contrast is also an important consideration. Objects that contrast sharply with the surrounding environment are more easily identified and focused on, and thus their weights may be correspondingly increased. Meanwhile, the system also examines the association degree of the object and other elements in the image to determine the importance and influence of the object in engineering management. Combining the above factors, the system calculates a specific enhanced rendering confidence weight for each original enhanced presentation tracking object. The weight not only represents the influence coefficient of the object in the panoramic image, but also provides an important basis for subsequent image rendering and optimization. For example, in a panoramic image containing multiple architectural structures, the system may calculate different enhanced rendering confidence weights for each structure based on factors such as its location, size, color contrast, and association with the surrounding environment. These weights will directly affect the visual effect of the various structures when the image is ultimately rendered, ensuring that important engineering management elements can get sufficient prominence and attention.

In another exemplary application scenario, a panoramic interaction processing system obtains a multi-mode image descriptor of an initial engineering management panoramic monitoring image. The descriptors are used as the representation of the image interaction element vectors, capture key characteristics such as colors, textures, shapes and the like in the images, and provide a rich information basis for subsequent image analysis and processing. Next, the panoramic interactive processing system uses these multimode image descriptors to accurately identify key elements in the initial engineering management panoramic monitoring image, namely the original enhanced presentation tracking object. These objects may include important equipment at the job site, building structures, or specific work areas, etc., that need to be particularly focused and tracked during engineering management. After determining these original enhanced presentation tracking objects, the panoramic interaction processing system further calculates enhanced rendering confidence weights for each object. This weight reflects not only the importance and influence of the object in the image, but also a number of factors such as the size, location, color contrast, and relevance to the surrounding environment of the object. Through complex algorithms and data analysis, the system can assign a reasonable weight value to each object. Finally, according to the enhanced rendering confidence weights, the panoramic interaction processing system generates an optimized indication of the panoramic monitoring image. These indications may include adjusting the display effect of the object, adding dynamic annotations, providing interactive information, etc., with the aim of improving the visual effect and information transfer efficiency of the panoramic image. The system applies different optimization strategies to different objects according to the weight, so that important engineering management elements can be fully highlighted and displayed in the panoramic image. Thus, through intelligent processing of the panoramic interaction processing system, the panoramic monitoring image of initial engineering management is comprehensively optimized and promoted. When the system is used, a user can more intuitively know the condition of a construction site and rapidly position and track key engineering management elements, so that the efficiency and accuracy of engineering management are improved. The whole process is completely driven based on data and an algorithm, and the objectivity and the accuracy of the processing are ensured.

According to the embodiment of the application, the multimode image descriptors of the initial engineering management panoramic monitoring image are obtained, so that the interactive elements of the image can be comprehensively and accurately captured, and a solid data base is provided for subsequent image processing. Furthermore, by determining the original enhanced display tracking object by using the descriptors, the key elements in the engineering management panoramic monitoring image can be identified and tracked, and the accuracy and efficiency of engineering management are greatly improved.

In addition, the embodiment of the application also quantifies the influence of the objects in the image by determining the enhanced rendering confidence weight of each original enhanced display tracking object, and provides scientific basis for subsequent image optimization. Finally, according to the panoramic monitoring image optimization indication determined by the weights, the visual effect and the information transmission capability of the panoramic image can be remarkably improved, so that engineering management personnel can acquire key information more intuitively and rapidly, and further accurate judgment and decision can be made. Therefore, the recognition precision, tracking efficiency and information presentation quality of the engineering management panoramic monitoring image can be improved, and the efficiency and the intelligent degree of engineering management are improved.

In some possible embodiments, the determining the enhanced rendering confidence weights for each of the original enhanced presentation tracking objects includes: acquiring pixel cluster wavelet characteristics of each original enhanced display tracking object; and determining the enhanced rendering confidence weight of each original enhanced display tracking object according to the pixel cluster wavelet characteristics of each original enhanced display tracking object and the multimode image descriptors of the initial engineering management panoramic monitoring image.

Based on this embodiment, determining the enhanced rendering confidence weights for each original enhanced presentation tracking object is a key step. This process first involves acquiring pixel cluster wavelet features of each original enhanced presentation tracking object. Pixel cluster wavelet features are a feature description method capable of capturing local texture and detail information of an image by analyzing wavelet transform coefficients of pixel clusters in different scales and directions to extract features. Panoramic interactive processing systems utilize advanced image processing techniques to wavelet transform clusters of pixels of these tracked objects to obtain their wavelet characteristics.

Next, the panoramic interaction processing system determines enhanced rendering confidence weights for each original enhanced presentation tracking object based on the pixel cluster wavelet features and the multi-mode image descriptors of the initial engineering management panoramic monitoring image. The multimode image descriptor in the embodiment of the application not only contains the global information of the image, but also reflects the relationship and characteristics among different areas in the image. The system can more comprehensively evaluate the importance and influence of each tracking object in the panoramic image by comprehensively considering wavelet characteristics and global descriptors.

In particular operations, the panoramic interactive processing system uses complex algorithms, such as machine learning algorithms, to analyze and compare these features and assign each tracked object a reinforced rendering confidence weight. This weight is in fact a numerical value representing the visual salience and the information importance of the object in the panoramic image. The higher the weight, the more the object needs to be rendered and highlighted in the panoramic image.

Therefore, the panoramic interaction processing system not only can accurately identify and track key elements in the engineering management panoramic monitoring image, but also can conduct personalized rendering and optimization according to the importance and characteristics of the key elements. The visual effect and information transfer efficiency of the panoramic image are greatly improved, so that engineering management staff can more intuitively know the condition of a construction site, and rapidly locate and pay attention to key areas.

Summarizing, the panoramic interaction processing system realizes accurate identification and personalized rendering of key elements in engineering management panoramic monitoring images by comprehensively utilizing pixel cluster wavelet characteristics and global multimode image descriptors. The visual effect of the image is improved, the information expression capability of the image is enhanced, and great convenience and benefit are brought to engineering management work. The intelligent image processing method can certainly promote the engineering management industry to develop to a more efficient and more accurate direction.

In some examples, the multimode image descriptor of the initial engineering management panoramic monitoring image includes first linear embedded knowledge representing an image interaction element vector of the initial engineering management panoramic monitoring image global; the determining the enhanced rendering confidence weight of each original enhanced display tracking object according to the pixel cluster wavelet characteristics of each original enhanced display tracking object and the multimode image descriptors of the initial engineering management panoramic monitoring image comprises the following steps: and determining the enhanced rendering confidence weight of each original enhanced display tracking object according to the commonality evaluation between the pixel cluster wavelet characteristics of each original enhanced display tracking object and the first linear embedded knowledge.

Based on this embodiment, the multimode image descriptor of the initial engineering management panoramic monitoring image plays a vital role. The descriptor contains an element called first linear embedded knowledge, which is an advanced data representation form and can comprehensively and accurately describe the image interaction element vector of the image global. In other words, the first linear embedded knowledge appears to be a summarized tag that summarizes the most core, essential interaction features in the panoramic monitoring image.

When determining the enhanced rendering confidence weight of the original enhanced presentation tracking object, the panoramic interactive processing system adopts a refined method. First, the system extracts pixel cluster wavelet features for each original enhanced display tracking object, a feature representation that is capable of carefully delineating the local texture and structure of the image. The system then compares these wavelet features to the first linear embedded knowledge looking for commonalities between them.

This process of commonality assessment is effectively a process of similarity measure, and the system calculates the similarity or correlation between the wavelet features of each tracked object and the first linear embedded knowledge. Such similarity can be calculated by a variety of mathematical methods, such as cosine similarity, pearson correlation coefficient, etc., with the particular choice of which method depends on the design and actual requirements of the system.

Once these commonality assessment indicators are obtained, the panoramic interaction processing system uses them to determine enhanced rendering confidence weights for each original enhanced presentation tracking object. This weight is a result of a comprehensive consideration of the importance, saliency and similarity to global features of the tracked object in the image. The greater the weight means that the tracked object is more consistent with the global feature and therefore should be more emphasized and highlighted in subsequent image presentations.

In this way, the panoramic interactive processing system can ensure that tracked objects which are highly consistent with the global features are more obviously displayed in the image, thereby helping a user to capture key information in the image more quickly. This not only improves the visual effect of the image, but also greatly enhances the information transfer efficiency of the image.

In this way, the panoramic interactive processing system realizes accurate determination of the enhanced rendering confidence weight of the original enhanced display tracking object by utilizing the first linear embedded knowledge and the commonality evaluation of the wavelet characteristics of the pixel clusters. The method not only improves the visual presentation quality of the image, but also enables the key information to be more prominent and easy to identify, and brings remarkable convenience and benefit for application in the fields of engineering management and the like.

In an alternative embodiment, the determining the panoramic monitoring image optimization designation of the initial engineering management panoramic monitoring image according to the enhanced rendering confidence weights of the respective original enhanced display tracking objects includes: determining an original enhanced display tracking object with the enhanced rendering confidence weight meeting a first judging requirement as an enhanced display tracking object contained in the initial engineering management panoramic monitoring image; wherein the first determination requirement includes a minimum one of: the enhanced rendering confidence weight is not smaller than a first preset value; and sorting all the original enhanced display tracking objects according to the descending order of the enhanced rendering confidence weights, wherein p bits are positioned before queuing, and p is a positive integer.

In this embodiment, the process of determining the panoramic monitoring image optimization designation of the initial engineering management panoramic monitoring image is performed in accordance with the previously calculated enhanced rendering confidence weights for each of the original enhanced display tracking objects. In this process, a key step is to screen out those tracking objects that meet specific decision requirements for subsequent enhanced presentation.

Specifically, the panoramic interaction processing system first sets a first decision requirement, which may be based on a specific value or relative ordering of enhanced rendering confidence weights. For example, the system may set a first preset value, and only if the enhanced rendering confidence weight of an original enhanced presentation tracking object is not less than the preset value, the object is considered to meet the decision requirement. The purpose of this is to ensure that only those tracked objects that meet certain criteria for importance and impact will be selected for subsequent enhancement processing.

Another way of determining is based on ordering. The panoramic interactive processing system sorts each original enhanced presentation tracking object according to the descending order of enhanced rendering confidence weights, and then selects p-bit objects before queuing, wherein p is a positive integer. This way it can be ensured that the selected tracked object is the most weighted, i.e. the most needed highlighted part of the image.

Once the original enhanced presentation tracking objects are selected that meet the first decision requirement, the panoramic interaction processing system determines them as enhanced presentation tracking objects contained in the initial engineering management panoramic monitoring image. These selected objects will be of particular interest and enhancement in subsequent image optimization processes to enhance the visual effect and information transfer capabilities of the image.

In this way, the panoramic interactive processing system can accurately identify and select the most critical and important parts in the image, and perform targeted optimization processing on the parts. This not only helps to improve the overall quality of the image, but also ensures that critical information is adequately presented and emphasized in the image.

Therefore, the panoramic interactive processing system screens and determines the enhanced display tracking object according to the enhanced rendering confidence weight by setting a clear first judging requirement, so that the optimization processing of the initial engineering management panoramic monitoring image is realized. The processing method not only improves the visual effect of the image, but also ensures accurate transmission of key information, and provides powerful technical support for application in the fields of engineering management and the like.

In another alternative embodiment, the determining the panoramic monitoring image optimization designation of the initial engineering management panoramic monitoring image according to the enhanced rendering confidence weights of the respective original enhanced display tracking objects includes: determining the original enhanced display tracking object with the enhanced rendering confidence weight meeting the second judging requirement as a key enhanced display tracking object contained in the initial engineering management panoramic monitoring image; wherein the second determination requirement includes a minimum one of: the enhanced rendering confidence weight is not smaller than a second preset value; and sorting all the original enhanced display tracking objects according to the descending order of the enhanced rendering confidence weights, wherein q bits are positioned before queuing, and q is a positive integer.

Based on this embodiment, the manner of determining the panoramic monitoring image optimization indication is more focused on identifying and highlighting key information in the image. The core of this process is to determine which objects are key elements in the image based on the enhanced rendering confidence weights of the respective original enhanced presentation tracking objects, and to subject them to special enhancement processing.

Specifically, the system first calculates an enhanced rendering confidence weight for each original enhanced presentation tracking object, which reflects the importance and impact of the object in the image. Then, the system sets a second decision requirement for screening out those objects with higher weights, i.e. key elements in the image.

The second decision requirement may take two forms. One is to set a specific second preset value, and only when the enhanced rendering confidence weight of a certain object is not smaller than the preset value, it is determined as a key enhanced presentation tracking object. This way it is ensured that the selected object has a significant importance and influence in the image.

Another way of determining is based on ordering. The panoramic interactive processing system sorts the objects according to the descending order of the enhanced rendering confidence weights, and then selects the q-bit objects before queuing as key enhanced display tracking objects, wherein q is a positive integer. This way it is ensured that the selected object is the most weighted, i.e. the most critical, most needed to be highlighted, part of the image.

Once the key enhanced presentation tracking objects are determined, the panoramic interactive processing system performs special enhancement processing on them. Such processing may include increasing the brightness, contrast, color saturation, etc. of these objects to make them more noticeable and noticeable in the image. At the same time, the system may also add special visual effects to these objects, such as halos, borders, etc., to further emphasize their importance.

In this way, the panoramic interactive processing system can accurately identify and highlight key elements in the image, enabling the user to capture such information faster and make accurate decisions and decisions. The method has very important application value in the fields of engineering management and the like, and can help a user to quickly find potential problems and risk points, so that the working efficiency and accuracy are improved.

Therefore, the panoramic interactive processing system determines the key enhanced display tracking object according to the enhanced rendering confidence weight by setting a reasonable second judging requirement, and realizes accurate identification and highlighting of key information in the image. The processing method not only improves the visual effect of the image, but also greatly improves the information transmission efficiency and accuracy of the image, and brings remarkable convenience and benefit for the application in various fields.

In some optional embodiments, the determining at least one original enhanced presentation tracking object in the initial engineering management panoramic monitoring image based on the multimode image descriptor of the initial engineering management panoramic monitoring image comprises: determining monitoring event label information of the initial engineering management panoramic monitoring image based on a multimode image descriptor of the initial engineering management panoramic monitoring image, wherein the monitoring event label information of the initial engineering management panoramic monitoring image is used for representing labels of monitoring events of the initial engineering management panoramic monitoring image; and determining at least one original enhanced display tracking object in the initial engineering management panoramic monitoring image based on the multimode image descriptor of the initial engineering management panoramic monitoring image and monitoring event label information of the initial engineering management panoramic monitoring image.

Based on this embodiment, the panoramic interaction processing system determines at least one original enhanced presentation tracking object in the initial engineering management panoramic monitoring image by a specific algorithm and procedure.

First, the panoramic interaction processing system determines monitoring event tag information for a panoramic monitoring image based on a multimodal image descriptor of the initial project management image. The multimode image descriptor in the embodiment of the application can be understood as the characteristic description of the image, and the multimode image descriptor contains various information such as color, texture, shape and the like of the image, thereby being beneficial to the comprehensive understanding of the image content by a system. The "monitoring event label information" refers to the category or property of the monitored event in the image, such as building construction, equipment failure, etc.

The system carries out deep analysis on the panoramic monitoring image of the initial engineering management through an image recognition technology, extracts key information from the panoramic monitoring image, and marks corresponding labels on the monitoring events. These tags not only help the system understand the type of event in the image, but also provide an important reference basis for subsequent processing.

Next, the panoramic interaction processing system combines the multimodal image descriptor of the initial engineering management panoramic monitoring image and the monitoring event tag information to determine at least one original enhanced presentation tracking object in the image. The original enhanced display tracking object in the embodiment of the application refers to a target which needs to be focused and tracked in an image, such as specific equipment, personnel or important structures in a construction site.

The system accurately identifies and locates the objects to be displayed and tracked in an enhanced manner through a series of complex algorithms according to image features and monitoring event tag information provided by the multimode image descriptors. The process involves various computer vision techniques such as image segmentation, target recognition, feature matching, etc., ensuring that the system can accurately capture key information in the image.

Finally, the panoramic interactive processing system continuously tracks and monitors the original enhanced display tracking objects and grasps the dynamic changes of the original enhanced display tracking objects in real time. The method is not only helpful for improving the efficiency and accuracy of engineering management, but also can timely discover and deal with potential problems and risks.

In summary, through intelligent analysis and processing of the panoramic interaction processing system, key objects can be accurately identified and tracked from the initial engineering management panoramic monitoring image, and more comprehensive and accurate data support is provided for engineering management. The technical scheme not only improves the intelligent level of engineering management, but also greatly improves the working efficiency and the safety, and brings revolutionary reform to the engineering management industry.

Under some preferred design ideas, the determining the monitoring event tag information of the initial engineering management panoramic monitoring image based on the multimode image descriptor of the initial engineering management panoramic monitoring image includes: determining discrimination possibility corresponding to each of X monitoring event labels based on the multimode image descriptors of the initial engineering management panoramic monitoring image; the judging possibility corresponding to a u-th monitoring event tag in the X monitoring event tags is used for indicating the confidence that the initial engineering management panoramic monitoring image belongs to the u-th monitoring event tag, X is an integer greater than 1, and u is a positive integer not greater than X; based on the discrimination possibility respectively corresponding to the X monitoring event labels, selecting v monitoring event labels with the maximum discrimination possibility from the X monitoring event labels to obtain v target monitoring event labels, wherein v is a positive integer; based on the discrimination possibility corresponding to the v target monitoring event labels, splicing the mapping characteristics corresponding to the v target monitoring event labels respectively to obtain mapping splicing characteristics; the monitoring event label information of the initial engineering management panoramic monitoring image comprises the mapping splicing characteristic.

Based on the design thought, the panoramic interaction processing system determines monitoring event label information of the initial engineering management panoramic monitoring image through a series of precise steps.

Firstly, the panoramic interaction processing system determines discrimination possibilities corresponding to X monitoring event labels respectively based on multimode image descriptors of initial engineering management panoramic monitoring images. The multimode image descriptor in the embodiment of the application is a comprehensive characteristic representation of the image, and integrates various information of the image, such as color, texture, shape and the like, so that the system can more comprehensively and accurately understand the content of the image. The "discrimination possibility" refers to the confidence that the image belongs to a specific monitoring event label, that is, the matching degree of the image and the label.

Specifically, the system analyzes each monitoring event label one by one, and evaluates the association degree of each monitoring event label according to the multimode image descriptor of the image, so as to obtain a value of discrimination possibility. The higher this value, the higher the degree of matching of the image to the tag.

Next, the panoramic interaction processing system selects v tags having the highest discrimination possibility from the X monitoring event tags, and these selected tags are referred to as "target monitoring event tags". The purpose of this step is to screen out the most relevant, best matching tags to the image content for subsequent processing and analysis.

After the target monitoring event labels are selected, the panoramic interaction processing system further splices the mapping features corresponding to the labels based on the discrimination possibility corresponding to the labels. "mapping features" in embodiments of the present application may be understood as specific image features associated with each tag that reflect the specific content of the image associated with the tag. By stitching these map features, the system can obtain a more comprehensive and rich representation of the image features, i.e., the "map stitching features".

Finally, the panoramic interaction processing system takes the mapping splice characteristics as a part of monitoring event label information of the initial engineering management panoramic monitoring image. Not only does this information help the system understand the image content more deeply, it also provides powerful data support for subsequent image processing, analysis, and applications.

Through the implementation mode, the panoramic interaction processing system can realize accurate analysis and processing of engineering management panoramic monitoring images. The beneficial effects are that: the method can improve the accuracy and efficiency of image processing, and help a user to acquire key information in an image more quickly; meanwhile, by means of the mode of mapping the splicing characteristics, the system can generate more comprehensive and detailed image description, and a richer data basis is provided for subsequent application. In general, the technical scheme can remarkably improve the processing effect and the application value of engineering management panoramic monitoring images.

Under other preferred design ideas, the multimode image descriptor of the initial engineering management panoramic monitoring image comprises linear embedded knowledge corresponding to a plurality of image blocks in the initial engineering management panoramic monitoring image respectively; the determining at least one original enhanced display tracking object in the initial engineering management panoramic monitoring image based on the multimode image descriptor of the initial engineering management panoramic monitoring image and the monitoring event label information of the initial engineering management panoramic monitoring image comprises the following steps: combining the linear embedded knowledge corresponding to each image block in the initial engineering management panoramic monitoring image with monitoring event label information of the initial engineering management panoramic monitoring image to obtain linear embedded combined knowledge corresponding to each image block; based on the linear embedded combined knowledge, at least one original enhanced presentation tracking object in the initial engineering management panoramic monitoring image is determined.

Under the design thought, the panoramic interaction processing system utilizes multimode image descriptors of the initial engineering management panoramic monitoring image and monitoring event label information to determine at least one original enhanced display tracking object in the image. The detailed implementation of this process is as follows: firstly, it should be clear that the "multimode image descriptor" in the embodiment of the present application includes linear embedded knowledge corresponding to each of a plurality of image blocks in the initial engineering management panoramic monitoring image. By "linear embedded knowledge" is understood a mathematical representation of the image block obtained by feature extraction and encoding that effectively captures the essential features of the image block for subsequent analysis and processing.

And then, the panoramic interaction processing system respectively combines the linear embedded knowledge corresponding to each image block in the initial engineering management panoramic monitoring image with the monitoring event tag information. This process is implemented by a specific algorithm, which aims to fuse the features of the image blocks with the corresponding monitoring event tag information, so as to obtain the corresponding 'linear embedded combination knowledge' of each image block. The combined knowledge contains characteristic information of the image blocks and tag information related to the monitoring event, so that the combined knowledge is more representative and distinguishable.

After obtaining the linear embedded combined knowledge corresponding to each image block, the panoramic interaction processing system further utilizes the information to determine at least one original enhanced presentation tracking object in the initial engineering management panoramic monitoring image. Specifically, the system identifies regions in the image with significant features or specific patterns, which are considered objects that require enhanced presentation and tracking, based on feature distribution and similarity measures of the linear embedded combined knowledge.

As such, not only relies on advanced image processing techniques and feature extraction methods, but also needs to be implemented with efficient algorithms and computational resources. Through the comprehensive application of the technical means, the panoramic interaction processing system can accurately identify key objects from complex engineering management panoramic monitoring images, and powerful support is provided for subsequent analysis, decision and display.

In other words, by fusing the linear embedded knowledge of the image blocks and monitoring event label information, the accuracy and reliability of object identification are improved; meanwhile, object tracking and display enhancement are performed by utilizing linear embedded combined knowledge, so that key information is more prominent and easy to understand. The method is not only beneficial to improving the efficiency and quality of engineering management, but also provides a beneficial technical support for intelligent application in the related field.

In some alternative embodiments, the panoramic monitoring image optimization indication of the initial engineering management panoramic monitoring image is determined by a depth residual memory network comprising an image description mining branch, a tracking object location branch, and an enhanced rendering processing branch; the image description mining branch is used for acquiring multimode image descriptors of the initial engineering management panoramic monitoring image; the tracking object positioning branch is used for determining at least one original enhanced display tracking object in the initial engineering management panoramic monitoring image based on a multimode image descriptor of the initial engineering management panoramic monitoring image; the enhanced rendering processing branch is used for determining enhanced rendering confidence weights of the original enhanced display tracking objects.

Based on this embodiment, the panoramic interaction processing system utilizes a depth residual memory network to determine a panoramic monitoring image optimization indication for the initial engineering management panoramic monitoring image. This depth residual memory network is a complex deep learning model that includes three main branches: the image description mines branches, tracks object locating branches, and strengthens rendering processing branches, each with its specific functions and tasks.

First, the main task of the image description mining branch is to acquire a multimode image descriptor of an initial engineering management panoramic monitoring image. The multimode image descriptor is a data structure capable of comprehensively describing image characteristics and comprises various information such as colors, textures, shapes and the like of images. The branch carries out deep analysis and processing on the image through a deep learning technology, thereby extracting the key characteristic information.

The following is the tracking object locating branch. The responsibility of this branch is to determine at least one original enhanced presentation tracking object in the image based on the initial engineering management panoramic monitoring image multimode image descriptor. In short, the targets that need to be focused and tracked in the image, such as equipment, personnel, etc. at the construction site, are to be found. This branch utilizes advanced object detection algorithms and deep learning techniques to accurately identify and locate these key objects.

And finally, strengthening rendering processing branches. The main goal of this branch is to determine the enhanced rendering confidence weights for the respective original enhanced presentation tracking objects. Confidence weights are a number that represents the importance of objects and rendering priority, which determines which objects should be more prominently shown in the panoramic image. This branch assigns a reasonable confidence weight to each subject by comprehensively considering the characteristics, position, size, etc. of the subject, as well as the overall composition and visual effect of the panoramic image.

The whole depth residual memory network can realize comprehensive optimization of the initial engineering management panoramic monitoring image through the cooperative work of the three branches. The technical scheme has the beneficial effects that: by applying the deep learning technology, the intelligent level and the automation degree of image processing are improved; meanwhile, through a multi-branch network structure, deep mining of image features and accurate positioning of objects are realized; and finally, the key information in the panoramic image is more prominent and easy to identify by strengthening rendering treatment, so that the efficiency and accuracy of engineering management are improved.

In other exemplary embodiments, the method for debugging the depth residual memory network includes: acquiring a multimode image descriptor of an engineering management panoramic monitoring image example through the image description mining branch, wherein the multimode image descriptor of the engineering management panoramic monitoring image example is used for representing an image interaction element vector of the engineering management panoramic monitoring image example; determining, by the tracking object positioning branch, an enhanced presentation tracking object recognition result of the engineering management panoramic monitoring image example based on a multimode image descriptor of the engineering management panoramic monitoring image example, the enhanced presentation tracking object recognition result including at least one original enhanced presentation tracking object in the engineering management panoramic monitoring image example; determining enhanced rendering confidence weights of the original enhanced display tracking objects through the enhanced rendering processing branches, wherein the enhanced rendering confidence weights are used for representing influence coefficients of the original enhanced display tracking objects in the engineering management panoramic monitoring image example; and determining a key tracking object identification result of the engineering management panoramic monitoring image example according to the enhanced rendering confidence weights of the original enhanced display tracking objects, wherein the key tracking object identification result comprises the following steps: at least one key enhanced presentation tracking object determined from the at least one original enhanced presentation tracking object based on the enhanced rendering confidence weight; and debugging the depth residual memory network based on the enhanced display tracking object recognition result, the key tracking object recognition result, the enhanced display tracking object priori basis and the key tracking object priori basis of the engineering management panoramic monitoring image example.

Based on the embodiment, the panoramic interaction processing system elaborates the debugging method of the depth residual memory network. This commissioning process is to ensure that the network can accurately identify and enhance the presentation of key objects in the engineering management panoramic monitoring image.

Firstly, the panoramic interaction processing system acquires multimode image descriptors of engineering management panoramic monitoring image examples through image description mining branches. These descriptors are in fact image interaction element vectors that are able to describe the various features in the image, such as color, texture, shape, etc., as well as the relationships between these features, in a comprehensive and detailed manner. This is the basis for understanding the image content and performing subsequent processing.

Next, a tracking object location branch may determine enhanced presentation tracking object recognition results for the engineering management panoramic monitoring image example based on the multimode image descriptors. In this process, the system identifies at least one original enhanced presentation tracking object in the image, which are parts that require special attention and enhancement in subsequent processing.

The enhanced rendering processing branch then determines an enhanced rendering confidence weight for each original enhanced presentation tracking object. This weight reflects the importance and influence of the object in the image, the greater the weight, the more important the object in the image, which needs to be shown more prominently.

After determining the enhanced rendering confidence weights for each object, the system determines key tracking object recognition results for the engineering management panoramic monitoring image example according to the weights. Specifically, the system screens at least one key enhanced presentation tracking object from at least one original enhanced presentation tracking object based on the enhanced rendering confidence weight. These key objects are the most important, most interesting parts of the image.

And finally, the panoramic interaction processing system utilizes the enhanced display tracking object recognition result and the key tracking object recognition result of the engineering management panoramic monitoring image example, and the preset enhanced display tracking object priori basis and the key tracking object priori basis to debug the depth residual memory network. The aim of debugging is to optimize parameters and structures of the network, so that the parameters and structures can more accurately identify key objects in the image and give more reasonable enhanced rendering confidence weights.

Through the series of debugging steps, the panoramic interaction processing system can ensure that the depth residual memory network plays an optimal role in practical application. The debugging method not only improves the accuracy of network identification, but also enhances the capturing and displaying capability of the system on key information, thereby providing more reliable and efficient technical support for engineering management.

In the following step, the debugging the depth residual memory network based on the enhanced display tracking object recognition result, the key tracking object recognition result, the enhanced display tracking object prior basis and the key tracking object prior basis of the engineering management panoramic monitoring image example comprises the following steps: determining a first training error based on the enhanced display tracking object recognition result and the enhanced display tracking object prior basis, wherein the first training error is used for evaluating the quality scores of the tracking object positioning branches; determining a second training error based on the key tracking object recognition result and the key tracking object prior basis, wherein the second training error is used for evaluating the quality score of the enhanced rendering processing branch; and debugging the depth residual memory network based on the first training error and the second training error.

In detail, the panoramic interaction processing system performs fine debugging on the depth residual memory network according to the enhanced display tracking object recognition result and the key tracking object recognition result of the engineering management panoramic monitoring image example obtained before, as well as a preset enhanced display tracking object priori basis and a preset key tracking object priori basis.

First, the system determines a first training error based on the enhanced presentation tracking object recognition result and the enhanced presentation tracking object a priori basis. As used herein, the term "enhanced presentation tracking object recognition result" refers to the original enhanced presentation tracking object obtained by the tracking object positioning branch, and the term "enhanced presentation tracking object a priori basis" refers to some preset standard or expected value about these tracking objects. By comparing the differences between the two, the system can calculate a first training error which is mainly used to evaluate the quality score of the tracking object locating branch, that is, it can reflect the accuracy and reliability of the tracking object locating branch in identifying and locating the tracking object.

Second, the system determines a second training error based on the key tracked object recognition result and the key tracked object a priori basis. Similar to the first training error, the "key tracking object recognition result" is a key enhanced presentation tracking object obtained by enhancing the rendering processing branch, and the "key tracking object prior basis" is some preset standard or expected value about the key objects. By comparing the two, the system can derive a second training error that is primarily used to evaluate the quality score of the enhanced rendering branch, which characterizes the accuracy of the branch in determining key tracked objects and assigning enhanced rendering confidence weights.

Finally, comprehensively considering the first training error and the second training error by the panoramic interactive processing system, and comprehensively debugging the depth residual memory network. In the debugging process, the system adjusts the parameters and the structure of the network according to the two errors so as to improve the performance and the accuracy of the network. By the method, the system can ensure that the depth residual memory network can better identify and enhance the key objects in the engineering management panoramic monitoring image in practical application.

In conclusion, the debugging method is not only beneficial to improving the identification accuracy and reliability of the depth residual memory network, but also enables the panoramic interaction processing system to be more efficient and accurate when processing engineering management panoramic monitoring images. By continuously optimizing the network performance, the panoramic interaction processing system provides more powerful and intelligent technical support for engineering management.

In some independent embodiments, the enhanced presentation tracking object recognition results include likelihood training recognition results corresponding to respective image blocks in the engineering management panoramic monitoring image example, the likelihood training recognition results including training prediction confidence of the image blocks with respect to a plurality of object categories; the enhanced display tracking object prior basis comprises likelihood prior knowledge corresponding to each image block in the engineering management panoramic monitoring image example, wherein the likelihood prior knowledge comprises prior confidence degrees of the image blocks relative to a plurality of object types; the determining a first training error based on the enhanced presentation tracking object recognition result and the enhanced presentation tracking object prior basis includes: and determining the first training error based on the likelihood training recognition results of the respective corresponding image blocks in the engineering management panoramic monitoring image example and the likelihood priori knowledge of the respective corresponding image blocks in the engineering management panoramic monitoring image example.

Based on the embodiment, when the panoramic interactive processing system is used for debugging the depth residual memory network, the processing of enhancing the recognition result of the display tracking object and enhancing the prior basis of the display tracking object can be involved. Both of which are done for each image block in the engineering management panoramic monitoring image example.

First, the enhanced display tracking object recognition result is obtained by analyzing engineering management panoramic monitoring image examples through a tracking object positioning branch. The result comprises the probability training recognition result corresponding to each image block in the image. The term "likelihood training recognition result" herein refers to, for each image block, to which of a plurality of object categories it belongs, the system predicting and giving a training prediction confidence. This confidence reflects the confidence and certainty that the system belongs to a certain object class for that image block.

Likewise, the enhanced presentation tracking object a priori basis is also for each image block. It contains a priori confidence, i.e. "likelihood a priori knowledge", of the image block with respect to the various object classes. These prior knowledge are preset or derived based on extensive statistical analysis of the data, which represents a preliminary judgment that an image block belongs to a certain object class without current image information.

And when the first training error is determined, comprehensively considering the recognition result of the enhanced display tracking object and the prior basis of the enhanced display tracking object by the panoramic interaction processing system. Specifically, the system compares the likelihood training recognition results (i.e., training prediction confidence) and likelihood prior knowledge (i.e., prior confidence) for each image block. By calculating the difference or gap between the two, the system can derive a first training error. This error reflects the accuracy and reliability of tracking the object locating branch in identifying and classifying image blocks, the smaller the error, the better the performance of the branch.

The first training error determined in this way not only considers the actual recognition result of the system, but also combines prior knowledge, so that the evaluation is more comprehensive and objective. The panoramic interaction processing system can utilize the error to pertinently optimize and adjust the parameters and the structure of the tracking object positioning branch, thereby improving the performance and the accuracy of the tracking object positioning branch in engineering management panoramic monitoring image processing.

The method for determining the training error by combining the recognition result and the priori knowledge is helpful for the panoramic interactive processing system to evaluate the performance of the depth residual memory network more accurately and provides powerful basis for subsequent debugging and optimization. The intelligent level of the system is improved, and more efficient and accurate technical support is provided for engineering management.

In other independent embodiments, the critical tracking object prior exposure tracking object comprises at least one prior critical enhancement in the engineering management panoramic monitoring image example; the determining a second training error based on the key tracking object recognition result and the key tracking object prior basis includes: and determining the second training error based on the enhanced rendering confidence weights respectively corresponding to the original enhanced display tracking objects in the engineering management panoramic monitoring image example and the priori critical enhanced display tracking objects in the original enhanced display tracking objects.

Based on the embodiment, the panoramic interaction processing system depends on the key tracking object recognition result and the key tracking object prior basis when determining the second training error. These two elements together form the basis for evaluating the performance of the enhanced rendering process branches.

First, the critical tracking object a priori basis is an important reference point. It includes at least one a priori key enhancement presentation tracking object in the engineering management panoramic monitoring image example. By "a priori" is meant that these key objects are determined based on past experience, knowledge or preset criteria, which play a vital role in the image. These a priori key enhanced presentation tracking objects are targets that the system expects to accurately identify and emphasize in the key tracking object recognition results.

Next, the panoramic interaction processing system looks at the key tracking object recognition results. This result is derived by the enhanced rendering processing branch, which contains the individual original enhanced presentation tracking objects in the image and their corresponding enhanced rendering confidence weights. The enhanced rendering confidence weight is an important indicator that reflects the degree of importance of each original enhanced presentation tracking object in the image. The higher the weight, the more critical the object, and more attention and emphasis should be given.

When determining the second training error, the system compares the original enhanced display tracking objects in the key tracking object recognition result with their corresponding enhanced rendering confidence weights, and the prior key enhanced display tracking objects in the prior basis of the key tracking objects. Specifically, the system checks whether the recognition results accurately recognize a priori key enhanced presentation tracking objects and assign them a relatively high enhanced rendering confidence weight.

If the recognition result is found to have missing or wrongly recognized prior key enhanced display tracking objects or the corresponding enhanced rendering confidence weights are inconsistent with expectations, the system calculates a second training error. This error reflects the accuracy and reliability of the enhanced rendering processing branch in determining key tracked objects and assigning weights.

By the aid of the second training error determined by the method, parameters and structures of the enhanced rendering processing branches can be optimized and adjusted in a targeted manner by the panoramic interaction processing system, so that performance and accuracy of the panoramic interaction processing system in engineering management panoramic monitoring image processing are improved. This not only helps the system more accurately identify and emphasize key objects in the image, but also provides more reliable and efficient technical support for engineering management.

In summary, the method for determining the second training error based on the key tracking object recognition result and the key tracking object priori basis is an important means for improving the image processing capability and the intelligent level of the panoramic interactive processing system. By continuous optimization and adjustment, the system can more accurately identify and render key information in the image, thereby providing more comprehensive and accurate data support for engineering management.

In yet other independent embodiments, the depth residual memorization network further comprises a monitoring event decision branch, the method further comprising: determining a monitoring event tag identification result of the engineering management panoramic monitoring image example based on a multimode image descriptor of the engineering management panoramic monitoring image example through the monitoring event decision branch, wherein the monitoring event tag identification result of the engineering management panoramic monitoring image example comprises discrimination possibility prediction results corresponding to X monitoring event tags respectively; the judging possibility prediction result corresponding to a u-th monitoring event tag in the X monitoring event tags is used for indicating an initial probability value of the engineering management panoramic monitoring image example belonging to the u-th monitoring event tag, X is an integer greater than 1, and u is a positive integer not greater than X; determining monitoring event label information of the engineering management panoramic monitoring image example based on a monitoring event label identification result of the engineering management panoramic monitoring image example, wherein the monitoring event label information of the engineering management panoramic monitoring image example is used for representing a label to which a monitoring event of the engineering management panoramic monitoring image example belongs; the tracking object positioning branch is used for determining an enhanced display tracking object recognition result of the engineering management panoramic monitoring image example based on the multimode image descriptor of the engineering management panoramic monitoring image example and monitoring event label information of the engineering management panoramic monitoring image example; and determining a third training error based on the monitoring event label recognition result and the monitoring event label priori basis, wherein the third training error is used for evaluating the quality scores of the monitoring event decision branches. The debugging of the depth residual memory network based on the first training error and the second training error comprises: and debugging the depth residual memory network based on the first training error, the second training error and the third training error.

Based on the embodiment, the depth residual memory network of the panoramic interactive processing system is further expanded, and a component named as a decision branch of the monitoring event is added. The introduction of the new branch enables the system to analyze engineering management panoramic monitoring images more comprehensively and extract more abundant information from the engineering management panoramic monitoring images.

First, the role of the monitoring event decision branch is to determine the monitoring event tag recognition result of an engineering management panoramic monitoring image based on the multimode image descriptor of the image. The term "multimode image descriptor" as used herein refers to a comprehensive description of various features (such as color, texture, shape, etc.) in an image, while the term "monitoring event tag recognition result" refers to the category of monitoring event to which the image may belong. Specifically, the system generates a discriminatory likelihood prediction result for each possible monitored event tag, which is in effect an initial probability value indicating the likelihood that the image belongs to a particular monitored event tag.

Once the monitoring event tag identification result is obtained, the panoramic interaction processing system determines monitoring event tag information of the engineering management panoramic monitoring image example according to the monitoring event tag information. This tag information will ultimately be used to describe the category to which a particular monitored event occurs in the image. Notably, the tracking object location branch uses this tag information in subsequent processing in conjunction with the multimode image descriptor to more accurately determine enhanced presentation tracking object recognition results in the image.

Next, the system calculates a new training error, namely a third training error, by using the monitoring event tag recognition result and a preset monitoring event tag priori basis. This error is mainly used to evaluate the performance of the monitoring event decision branch, i.e. whether it can accurately assign the correct monitoring event label to the image.

Finally, the panoramic interaction processing system integrates the first training error (from the tracked object location branch), the second training error (from the enhanced rendering processing branch), and the newly introduced third training error (from the monitoring event decision branch) when debugging the depth residual memory network. By comprehensively analyzing the three errors, the system can more comprehensively understand the performance bottleneck of the network and perform targeted optimization and adjustment according to the performance bottleneck.

Therefore, by introducing the monitoring event decision branch and the corresponding third training error, the panoramic interaction processing system can more accurately identify and position key objects in the image and can also allocate more accurate monitoring event labels for the objects when processing engineering management panoramic monitoring images. The intelligent level of the system is improved, and richer and accurate information support is provided for engineering management. Meanwhile, the comprehensive debugging method based on various training errors also enables the performance of the depth residual memory network to be more comprehensively optimized and improved.

In other independent embodiments, the prior basis of the monitoring event labels includes prior discrimination possibilities corresponding to the X monitoring event labels respectively, and the prior discrimination possibility corresponding to the u-th monitoring event label is used for indicating that the engineering management panoramic monitoring image example belongs to the prior probability value of the u-th monitoring event label; the determining a third training error based on the monitoring event tag identification result and the monitoring event tag prior basis includes: and determining the third training error based on the discrimination possibility prediction results respectively corresponding to the X monitoring event labels and the priori discrimination possibility respectively corresponding to the X monitoring event labels.

Based on this embodiment, the panoramic interaction processing system further refines the manner of use of the monitoring event tag a priori basis and how to use these a priori basis and monitoring event tag recognition results to determine the third training error.

Firstly, the prior basis of the monitoring event labels comprises prior discrimination possibilities corresponding to X monitoring event labels respectively. The "a priori likelihood of discrimination" is an important concept, and refers to the probability of occurrence of a certain monitoring event label preset according to past experience, statistical data or other reliable sources without any current image information. In other words, the a priori discrimination likelihood is a priori probability value that is used to represent the likelihood that the engineering management panoramic monitoring image instance belongs to a particular monitoring event tag.

Specifically, for each monitoring event tag, such as the u-th monitoring event tag, its corresponding a priori likelihood of discrimination indicates the probability that the tag will appear under normal conditions. This probability value may be used as an important reference for the panoramic interaction processing system when processing a new image.

Next, the panoramic interaction processing system uses the monitoring event tag recognition results and these a priori basis to determine a third training error. The identification result of the monitoring event tag is obtained by analyzing the current image by the system, and the identification result comprises the discrimination possibility prediction results corresponding to the X monitoring event tags respectively. These predictions are derived based on the characteristics and information of the current image, reflecting the predicted probability that the system belongs to each monitoring event tag for the current image.

In determining the third training error, the system compares the discrimination likelihood prediction result for each monitoring event tag to the corresponding a priori discrimination likelihood. By calculating the difference or gap between the two, the system can derive a training error value for the performance of the monitor event decision branch. This error value reflects the accuracy and reliability of the monitoring event decision branch in identifying and classifying the monitoring event tags.

In particular, if the discrimination likelihood prediction result of a certain monitoring event tag differs significantly from the corresponding a priori discrimination likelihood, then it is stated that the monitoring event decision branch may be problematic in processing the tag, requiring further optimization and adjustment. Conversely, if the two differ less, then the performance of the monitor event decision branch is illustrated as relatively good.

The third training error determined in the method not only considers the actual recognition result of the system, but also combines prior knowledge, so that the evaluation is more comprehensive and objective. The panoramic interaction processing system can utilize the error to pertinently optimize and adjust parameters and structures of the monitoring event decision branch, so that the performance and accuracy of the panoramic interaction processing system in engineering management panoramic monitoring image processing are improved.

Therefore, the method for determining the training error by combining the recognition result and the priori knowledge is an important means for improving the image processing capability and the intelligent level of the panoramic interactive processing system. Through continuous optimization and adjustment, the system can more accurately identify and classify the monitoring event labels, so that more comprehensive and accurate data support is provided for engineering management.

Further, fig. 2 is a schematic structural diagram of a panoramic interaction processing system 200 according to an embodiment of the present application. The panoramic interaction processing system 200 as shown in fig. 2 comprises a processor 210, from which the processor 210 may call and run a computer program to implement the method in an embodiment of the present application.

Optionally, as shown in fig. 2, the panoramic interaction processing system 200 may also include a memory 230. Wherein the processor 210 may call and run a computer program from the memory 230 to implement the method in an embodiment of the application.

Wherein the memory 230 may be a separate device from the processor 210 or may be integrated into the processor 210.

Optionally, as shown in fig. 2, the panorama interaction processing system 200 may further include a transceiver 220, and the processor 210 may control the transceiver 220 to interact with other devices, and in particular, may send information or data to other devices, or receive information or data sent by other devices.

Optionally, the panoramic interaction processing system 200 may implement the storage engine or a component (such as a processing module) in the storage engine or a corresponding flow corresponding to a device in which the storage engine is deployed in each method of the embodiments of the present application, which is not described herein for brevity.

It should be appreciated that the processor of an embodiment of the present application may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method embodiments may be implemented by integrated logic circuits of hardware in a processor or instructions in software form. The Processor may be a general purpose Processor, a digital signal Processor (DIGITAL SIGNAL Processor, DSP), an Application SPECIFIC INTEGRATED Circuit (ASIC), an off-the-shelf programmable gate array (Field Programmable GATE ARRAY, FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components. The disclosed methods, steps, and logic blocks in the embodiments of the present application may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present application may be embodied directly in the execution of a hardware decoding processor, or in the execution of a combination of hardware and software modules in a decoding processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in a memory, and the processor reads the information in the memory and, in combination with its hardware, performs the steps of the above method.

It will be appreciated that the memory in embodiments of the application may be volatile memory or nonvolatile memory, or may include both volatile and nonvolatile memory. The nonvolatile Memory may be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically Erasable EPROM (EEPROM), or a flash Memory. The volatile memory may be random access memory (Random Access Memory, RAM) which acts as external cache memory. By way of example, and not limitation, many forms of RAM are available, such as static random access memory (STATIC RAM, SRAM), dynamic random access memory (DYNAMIC RAM, DRAM), synchronous Dynamic Random Access Memory (SDRAM), double data rate Synchronous dynamic random access memory (Double DATA RATE SDRAM, DDR SDRAM), enhanced Synchronous dynamic random access memory (ENHANCED SDRAM, ESDRAM), synchronous link dynamic random access memory (SYNCHLINK DRAM, SLDRAM), and Direct memory bus RAM (DR RAM). It should be noted that the memory of the systems and methods described herein is intended to comprise, without being limited to, these and any other suitable types of memory.

It should be appreciated that the above memory is exemplary but not limiting, and for example, the memory in the embodiments of the present application may also be static random access memory (STATIC RAM, SRAM), dynamic random access memory (DYNAMIC RAM, DRAM), synchronous Dynamic Random Access Memory (SDRAM), double data rate synchronous dynamic random access memory (doubledata RATE SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (ENHANCED SDRAM, ESDRAM), synchronous link dynamic random access memory (SYNCH LINK DRAM, SLDRAM), direct Rambus RAM (DR RAM), and the like. That is, the memory in embodiments of the present application is intended to comprise, without being limited to, these and any other suitable types of memory.

On the basis of the above, a computer readable storage medium is provided, on which a computer program is stored, which computer program, when run, implements the method described above.

The foregoing is merely exemplary of the present application and is not intended to limit the present application. Various modifications and variations of the present application will be apparent to those skilled in the art.

Claims

1. A method for enhancing consultation technology by panoramic interactive presentation, which is applied to a panoramic interactive processing system, the method comprising: acquiring a multimode image descriptor of an initial engineering management panoramic monitoring image, wherein the multimode image descriptor of the initial engineering management panoramic monitoring image is used for representing an image interaction element vector of the initial engineering management panoramic monitoring image; determining at least one original enhanced presentation tracking object in the initial engineering management panoramic monitoring image based on a multimode image descriptor of the initial engineering management panoramic monitoring image; determining enhanced rendering confidence weights of the original enhanced display tracking objects, wherein the enhanced rendering confidence weights are used for representing influence coefficients of the original enhanced display tracking objects in the initial engineering management panoramic monitoring image; and determining panoramic monitoring image optimization instructions of the initial engineering management panoramic monitoring images according to the enhanced rendering confidence weights of the original enhanced display tracking objects.

2. The method of claim 1, wherein the determining the enhanced rendering confidence weights for each of the original enhanced presentation tracking objects comprises: acquiring pixel cluster wavelet characteristics of each original enhanced display tracking object; and determining the enhanced rendering confidence weight of each original enhanced display tracking object according to the pixel cluster wavelet characteristics of each original enhanced display tracking object and the multimode image descriptors of the initial engineering management panoramic monitoring image.

3. The method of claim 2, wherein the multimode image descriptor of the initial engineering management panorama monitoring image includes a first linear embedded knowledge representing an image interaction element vector of the initial engineering management panorama monitoring image global; the determining the enhanced rendering confidence weight of each original enhanced display tracking object according to the pixel cluster wavelet characteristics of each original enhanced display tracking object and the multimode image descriptors of the initial engineering management panoramic monitoring image comprises the following steps: and determining the enhanced rendering confidence weight of each original enhanced display tracking object according to the commonality evaluation between the pixel cluster wavelet characteristics of each original enhanced display tracking object and the first linear embedded knowledge.

4. The method of claim 1, wherein determining a panoramic monitoring image optimization designation for the initial engineering management panoramic monitoring image based on the enhanced rendering confidence weights for each of the original enhanced display tracked objects comprises: determining an original enhanced display tracking object with the enhanced rendering confidence weight meeting a first judging requirement as an enhanced display tracking object contained in the initial engineering management panoramic monitoring image; wherein the first determination requirement includes a minimum one of: the enhanced rendering confidence weight is not smaller than a first preset value; and sorting all the original enhanced display tracking objects according to the descending order of the enhanced rendering confidence weights, wherein p bits are positioned before queuing, and p is a positive integer.

5. The method of claim 1, wherein determining a panoramic monitoring image optimization designation for the initial engineering management panoramic monitoring image based on the enhanced rendering confidence weights for each of the original enhanced display tracked objects comprises: determining the original enhanced display tracking object with the enhanced rendering confidence weight meeting the second judging requirement as a key enhanced display tracking object contained in the initial engineering management panoramic monitoring image; wherein the second determination requirement includes a minimum one of: the enhanced rendering confidence weight is not smaller than a second preset value; and sorting all the original enhanced display tracking objects according to the descending order of the enhanced rendering confidence weights, wherein q bits are positioned before queuing, and q is a positive integer.

6. The method of claim 1, wherein the determining at least one original enhanced presentation tracking object in the initial engineering management panoramic monitoring image based on the multimode image descriptor of the initial engineering management panoramic monitoring image comprises: determining monitoring event label information of the initial engineering management panoramic monitoring image based on a multimode image descriptor of the initial engineering management panoramic monitoring image, wherein the monitoring event label information of the initial engineering management panoramic monitoring image is used for representing labels of monitoring events of the initial engineering management panoramic monitoring image; and determining at least one original enhanced display tracking object in the initial engineering management panoramic monitoring image based on the multimode image descriptor of the initial engineering management panoramic monitoring image and monitoring event label information of the initial engineering management panoramic monitoring image.

7. The method of claim 6, wherein the determining monitoring event tag information for the initial engineering management panoramic monitoring image based on the multimode image descriptor for the initial engineering management panoramic monitoring image comprises: determining discrimination possibility corresponding to each of X monitoring event labels based on the multimode image descriptors of the initial engineering management panoramic monitoring image; the judging possibility corresponding to a u-th monitoring event tag in the X monitoring event tags is used for indicating the confidence that the initial engineering management panoramic monitoring image belongs to the u-th monitoring event tag, X is an integer greater than 1, and u is a positive integer not greater than X; based on the discrimination possibility respectively corresponding to the X monitoring event labels, v monitoring event labels with the maximum discrimination possibility are selected from the X monitoring event labels, v target monitoring event labels are obtained, and v is a positive integer; based on the discrimination possibility corresponding to the v target monitoring event labels, splicing the mapping characteristics corresponding to the v target monitoring event labels respectively to obtain mapping splicing characteristics; the monitoring event label information of the initial engineering management panoramic monitoring image comprises the mapping splicing characteristic.

8. The method of claim 6, wherein the multimode image descriptor of the initial engineering management panoramic monitoring image includes linear embedded knowledge of respective correspondence of a plurality of image blocks in the initial engineering management panoramic monitoring image; the determining at least one original enhanced display tracking object in the initial engineering management panoramic monitoring image based on the multimode image descriptor of the initial engineering management panoramic monitoring image and the monitoring event label information of the initial engineering management panoramic monitoring image comprises the following steps: combining the linear embedded knowledge corresponding to each image block in the initial engineering management panoramic monitoring image with monitoring event label information of the initial engineering management panoramic monitoring image to obtain linear embedded combined knowledge corresponding to each image block; based on the linear embedded combined knowledge, at least one original enhanced presentation tracking object in the initial engineering management panoramic monitoring image is determined.

9. The method of any one of claims 1 to 8, wherein the panoramic monitoring image optimization indication of the initial engineering management panoramic monitoring image is determined by a depth residual memory network comprising an image description mining branch, a tracking object location branch, and an enhanced rendering processing branch; the image description mining branch is used for acquiring multimode image descriptors of the initial engineering management panoramic monitoring image; the tracking object positioning branch is used for determining at least one original enhanced display tracking object in the initial engineering management panoramic monitoring image based on a multimode image descriptor of the initial engineering management panoramic monitoring image; The enhanced rendering processing branch is used for determining enhanced rendering confidence weights of the original enhanced display tracking objects; the debugging method of the depth residual memory network comprises the following steps: acquiring a multimode image descriptor of an engineering management panoramic monitoring image example through the image description mining branch, wherein the multimode image descriptor of the engineering management panoramic monitoring image example is used for representing an image interaction element vector of the engineering management panoramic monitoring image example; determining, by the tracking object positioning branch, an enhanced presentation tracking object recognition result of the engineering management panoramic monitoring image example based on a multimode image descriptor of the engineering management panoramic monitoring image example, the enhanced presentation tracking object recognition result including at least one original enhanced presentation tracking object in the engineering management panoramic monitoring image example; Determining enhanced rendering confidence weights of the original enhanced display tracking objects through the enhanced rendering processing branches, wherein the enhanced rendering confidence weights are used for representing influence coefficients of the original enhanced display tracking objects in the engineering management panoramic monitoring image example; and determining a key tracking object identification result of the engineering management panoramic monitoring image example according to the enhanced rendering confidence weights of the original enhanced display tracking objects, wherein the key tracking object identification result comprises the following steps: at least one key enhanced presentation tracking object determined from the at least one original enhanced presentation tracking object based on the enhanced rendering confidence weight; Debugging the depth residual memory network based on the enhanced display tracking object recognition result, the key tracking object recognition result, the enhanced display tracking object priori basis and the key tracking object priori basis of the engineering management panoramic monitoring image example; the step of debugging the depth residual memory network based on the enhanced display tracking object recognition result, the key tracking object recognition result, the enhanced display tracking object priori basis and the key tracking object priori basis of the engineering management panoramic monitoring image example comprises the following steps: determining a first training error based on the enhanced display tracking object recognition result and the enhanced display tracking object prior basis, wherein the first training error is used for evaluating the quality scores of the tracking object positioning branches; determining a second training error based on the key tracking object recognition result and the key tracking object prior basis, wherein the second training error is used for evaluating the quality score of the enhanced rendering processing branch; and debugging the depth residual memory network based on the first training error and the second training error.

10. A panoramic interactive processing system comprising at least one processor and memory; the memory stores computer-executable instructions; the at least one processor executing computer-executable instructions stored in the memory causes the at least one processor to perform the method of any one of claims 1-9.