CN116664968A - Data processing method, model training method, device and electronic device - Google Patents

Abstract

The application provides a data processing method, a model training method, a device and electronic equipment, wherein the method comprises the following steps: obtaining a target data set of face images of a target object, wherein the face images in at least one image pair in the target data set are different, determining a target image pair with similarity smaller than a similarity threshold according to the similarity between the face images in at least one image pair, determining the occurrence frequency of each target face image according to the target face images included in the target image pair, and deleting any target face image from the target data set in response to the occurrence frequency of any target face image being greater than a quantity threshold. By determining the target image pair with poor similarity, counting the occurrence frequency of the target face image in the range of the target image pair, and carrying out data cleaning on the target data set based on the occurrence frequency, the face image data in the target data set are face images meeting the quality requirement, and the accuracy and efficiency are improved.

Description

Data processing method, model training method, device and electronic device

technical field

The present application relates to the field of computer technology, and in particular to a data processing method, a model training method, a device and electronic equipment.

Background technique

With the development of machine learning technology, such as face recognition technology based on machine learning has been widely used in all aspects of social life. Face recognition technology usually uses sample face images to train face recognition models, while training face recognition models based on deep learning usually requires massive amounts of face data.

In related technologies, the acquisition of massive data is usually crawled and saved from the Internet with keywords by crawlers. This type of data is characterized by a lot of noise and unbalanced data. Data cleaning is required to ensure that the image data in each data set is Data that meets the quality requirements, so how to improve the accuracy of data processing in the dataset is a technical problem that needs to be solved urgently.

Contents of the invention

This application aims to solve one of the technical problems in the related art at least to a certain extent.

For this reason, the present application proposes a data processing method, a model training method, a device, and electronic equipment, which improve the accuracy of data processing.

An embodiment of the present application proposes a data processing method, including:

Obtain a target data set of a face image of a target object; wherein, the target data set includes at least one image pair, and the face images in the image pair are different;

According to the similarity between the face images in the at least one image pair, determine a target image pair whose similarity is less than a similarity threshold;

According to the target human face images included in the target image pair, determine the frequency of occurrence of each of the target human face images;

In response to the occurrence frequency of any of the target human face images being greater than a quantity threshold, any of the target human face images is deleted from the target data set.

An embodiment of the present application proposes a model training method, including:

Obtain a sample target data set; wherein, the sample target data set contains a sample face image of the target object; the sample face image is marked with the identity information of the included face;

Inputting any of the sample human face images into the recognition model for classification to determine the identity information obtained by predicting any of the sample human face images;

The recognition model is trained based on the difference between the predicted identity information and the labeled identity information.

Another embodiment of the present application proposes a data processing device, including:

An acquisition module, configured to acquire a target data set of a face image of a target object; wherein, the target data set includes at least one image pair, and the face images in the image pair are different;

A first determination module, configured to determine a target image pair whose similarity is less than a similarity threshold according to the similarity between the face images in the at least one image pair;

The second determination module is used to determine the frequency of appearance of each target face image according to the target face image included in the target image pair;

A processing module, configured to delete any of the target face images from the target data set in response to the occurrence frequency of any of the target face images being greater than a quantity threshold.

Another embodiment of the present application proposes a model training device, including:

An acquisition module, configured to acquire a sample target data set; wherein, the sample target data set contains a sample face image of a target object; the sample face image is marked with the identity information of the included face;

A recognition module, configured to input any of the sample face images into a recognition model for classification, so as to determine the identity information obtained by predicting any of the sample face images;

A training module, configured to train the recognition model based on the difference between the predicted identity information and the labeled identity information.

Another embodiment of the present application proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and operable on the processor. When the processor executes the program, the aforementioned embodiments are implemented. the method described.

Another embodiment of the present application provides a non-transitory computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the methods described in the foregoing embodiments are implemented.

Another embodiment of the present application provides a computer program product, on which a computer program is stored, and when the program is executed by a processor, the method described in the foregoing embodiments is implemented.

The data processing method, model training method, device and electronic equipment proposed by the present application obtain a target data set of a face image of a target object, wherein the target data set includes at least one image pair, and the face images in the image pair are different, according to The similarity between the face images in at least one image pair, determine the target image pair whose similarity is less than the similarity threshold, determine the frequency of occurrence of each target face image according to the target face images included in the target image pair, and respond to If the frequency of occurrence of any target face image is greater than the number threshold, any target face image is deleted from the target data set. Through the similarity between face images, the target image pair with poor similarity is determined. Within the scope of the target image pair, the frequency of occurrence of the target face image in the target image pair is further counted, and the target image is calculated based on the frequency of occurrence. The data set is cleaned, so that the face image data in the target data set are all face images that meet the quality requirements, which improves the accuracy and efficiency.

Additional aspects and advantages of the application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the application.

Description of drawings

The above and/or additional aspects and advantages of the present application will become apparent and easy to understand from the following description of the embodiments in conjunction with the accompanying drawings, wherein:

FIG. 1 is a schematic flow diagram of a data processing method provided in an embodiment of the present application;

FIG. 2 is a schematic flow diagram of another data processing method provided in the embodiment of the present application;

FIG. 3 is a schematic flow diagram of another data processing method provided in the embodiment of the present application;

Fig. 4 is a schematic flow chart of another model training method provided by the embodiment of the present application;

FIG. 5 is a schematic structural diagram of a data processing device provided in an embodiment of the present application;

FIG. 6 is a schematic structural diagram of a model training device provided in an embodiment of the present application;

FIG. 7 is a block diagram of an electronic device provided by an embodiment of the present application.

Detailed ways

Embodiments of the present application are described in detail below, examples of which are shown in the drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the figures are exemplary, and are intended to explain the present application, and should not be construed as limiting the present application.

The following describes the data processing method, model training method, device, and electronic equipment of the embodiments of the present application with reference to the accompanying drawings.

In related technologies, the face images in the data set are cleaned to remove abnormal data or noise data, that is, through face clustering, the face images in each data set are clustered and the cluster centers are determined, and then the distance The image with a relatively far center is used to achieve the purpose of noise removal. This type of method is relatively simple and crude, and the processing accuracy is poor. At the same time, the number of cluster centers in the clustering algorithm itself cannot be determined, and the face data in different folders may have different numbers of cluster centers, which is easy to cause deviation, and this type of method has a large amount of calculation and low processing efficiency.

FIG. 1 is a schematic flowchart of a data processing method provided by an embodiment of the present application.

The execution subject of the data processing method in the embodiment of the present application is a data processing device, and the device may be set in an electronic device, where the electronic device may be a terminal device or a server, which is not limited in this embodiment.

As shown in Figure 1, the method may include the following steps:

Step 101, acquiring a target data set of a face image of a target object.

Among them, the target object is the target user.

In the embodiment of the present application, the face images contained in the target data set are face images of the target object, and at least one image pair is obtained by pairing the face images in the target data set, and the face images in the image pairs are different. As an implementation manner, the face images in the target data set are numbered, different face images correspond to different numbers, and two face images with different numbers are paired to obtain at least one image pair.

Step 102, according to the similarity between face images in at least one image pair, determine a target image pair whose similarity is smaller than a similarity threshold.

In the embodiment of the present application, for each image pair, the similarity between the face images in the image pair is determined, wherein the similarity indicates the similarity between the face images in the image pair, through which the similarity can be Indicates how likely the two face images are of the same object. The higher the similarity, the greater the possibility. Compare each similarity with the similarity threshold, that is, compare each similarity with the similarity threshold to determine The target image pairs whose similarity is less than the similarity threshold, that is, the target image pairs that do not meet the quality requirements are screened out.

Step 103, according to the target face images included in the target image pair, determine the occurrence frequency of each target face image.

In the embodiment of the present application, the target face images included in each target image pair are obtained, and the frequency of appearance of each face image is counted to obtain the frequency of appearance of each face image in all target image pairs. The frequency of appearance of the target face image indicates the frequency with which the similarity between the target face image and other face images in the target sample set is small. The higher the frequency, the greater the gap between the target face image and other face images in the target sample set, and the more likely it is an abnormal target face image.

As an example, there are 4 pairs of target images, which are respectively target image pair 1, target image pair 2, target image pair 3 and target image pair 4, wherein the number of the target face image contained in the target image pair 1 is 1 and 3, the number of the target face image contained in the target image pair 2 is 2 and 3, the number of the target face image contained in the target image pair 3 is 1 and 4, and the target person contained in the target image pair 4 The numbers of the face images are 1 and 2. Through statistics, the frequency of the target face image numbered 1 is 3 times, the frequency of the target face image numbered 2 is 2 times, and the frequency of the target face image numbered 3 is 3 times. The frequency of appearance is 2 times, and the frequency of appearance of the target face image numbered 4 is 1 time.

Step 104, deleting any target face image from the target data set in response to the occurrence frequency of any target face image being greater than the number threshold.

In the embodiment of the present application, the frequency of occurrence of each target face image is compared with the number threshold, and any target face image whose frequency of occurrence is determined to be greater than the number threshold can also be referred to as the first target face image for ease of identification. , then the first target face data is noise data or abnormal face data, then the first target face image is deleted from the target data set, which improves the accuracy of face data processing in the target data set. Further, the target data set obtained after processing can be used to train related face recognition models, and the accuracy of face recognition model training can be improved by improving the accuracy of face images used as training samples.

For example, if the number threshold is 2 times, the frequency of the target face image numbered 1 is 3 times, which is greater than the number threshold. Therefore, it is determined that the target face image numbered 1 is abnormal data and needs to be deleted from the target data set. To ensure that the face images of the target object in the target image set are all face images meeting the quality requirements.

It should be understood that the same process can be performed on the target image sets of other objects using the data processing method of the embodiment of the present application to obtain target image sets of other objects, so that a large number of image sets that meet the quality requirements can be obtained, and multiple processing The obtained image set can be used to train the recognition model to improve the training accuracy of the recognition model, wherein the recognition model obtained from face training is used for face recognition.

In the data processing method of the embodiment of the present application, the target data set of the face image of the target object is acquired, wherein the target data set includes at least one image pair, and the face images in the image pair are different, according to the person in the at least one image pair The similarity between the face images, determine the target image pair whose similarity is less than the similarity threshold, determine the frequency of occurrence of each target face image according to the target face image included in the target image pair, and respond to the appearance of any target face image If the frequency is greater than the number threshold, any target face image will be deleted from the target dataset. Through the similarity between face images, the target image pair with poor similarity is determined. Within the scope of the target image pair, the frequency of occurrence of the target face image in the target image pair is further counted, and the target image is calculated based on the frequency of occurrence. The data set is cleaned, so that the face image data in the target data set are all face images that meet the quality requirements, which improves the accuracy and efficiency.

Based on the above embodiment, Fig. 2 is a schematic flow diagram of another data processing method provided in the embodiment of the present application. As shown in Fig. 2, the method includes the following steps:

Step 201, acquiring an initial dataset of face images of a target object.

As an implementation method, the initial data set including the face image of the target object is obtained from the Internet through keywords, and the face image of the target object obtained from the network contains more noise data, or abnormal data, For example, the face image that does not belong to the target object, or, although it is the face image of the target object, the quality is poor and cannot be used, etc. Therefore, the initial data set needs to be cleaned to identify the noise data or the original data set. abnormal data.

Step 202, recognize the face images in the initial data set.

As an implementation manner, the face images in the initial data set may be detected through the face detection model, and whether each face image includes a face and the size of the included face can be identified.

Step 203: Delete the recognized face images whose size is smaller than the set threshold, or the face images that do not contain a face, from the initial data set to obtain the target data set.

As an implementation, for each face image in the initial data set, if the face image does not include a face, it is determined that the face image is noise data, and then deleted from the initial data set.

As another implementation, for each face image in the initial data set, if the face image includes a face, but the size of the face in the face image is smaller than the size threshold, or the proportion is smaller than the proportion threshold , it is determined that the face image is abnormal data, and then deleted from the initial data set, for example, the size threshold is 30*30 pixels, and the proportion threshold is one-tenth.

In the embodiment of the present application, by detecting the acquired initial data set to delete the size of the face or the image that does not contain the face to obtain the target data set, the initial cleaning process of the initial data set is realized, and the deleted Obvious abnormal data improves the effect of subsequent processing.

Step 204, acquiring a target dataset of face images of the target object.

Wherein, the explanations of the foregoing embodiments are also applicable to this embodiment, and the principle is the same, so details are not repeated here.

Step 205, for each image pair, determine the similarity between the face images in the image pair.

As an implementation, the face image in the image pair is input into the recognition model for feature extraction, and the image features of the face image in the image pair are obtained, and the relationship between the image features is determined according to the image features of the face image in the image pair. According to the distance between the image features, the similarity between the face images in the image pair is determined. Among them, the larger the distance between the image features of two face images in an image pair, the less similar the two face images in the image pair are, and the lower the similarity is; on the contrary, the two face images in an image pair The smaller the distance between the image features of the face images, the more similar the two face images in the image pair are, and the higher the similarity is.

Wherein, the distance between the image features may be cosine distance, Euclidean distance, etc., which can be used to determine the similarity between image features, which are not limited in this embodiment.

In step 206, each similarity is compared with a similarity threshold, and a target image pair whose similarity is smaller than the similarity threshold is determined.

Step 207, according to the target face images included in the target image pair, determine the occurrence frequency of each target face image.

Step 208, in response to the occurrence frequency of any target human face image being greater than the quantity threshold, deleting any target human face image from the target data set.

Wherein, for steps 206 to 208, reference may be made to the explanations in the foregoing embodiments, and the principles are the same, so details are not repeated here.

In the data processing method of the embodiment of the present application, the initial data set including the face image of the target object is obtained, and a part of obviously noise or abnormal data is removed through preliminary screening to obtain the target data set, which is further cleaned based on the target data set, That is, by calculating the distance between the face image features in each image pair, the similarity between the face images in the image pair is determined, and the target image is statistically obtained from the determined target image pair with poor similarity. The frequency of the target face image in the alignment, the higher the frequency, the greater the difference between the target face image and other face images in the target data, and the target face image may belong to noise data or abnormal data The greater the accuracy, further, the noise data cleaning is performed on the target data set based on the frequency of occurrence, so that the face image data in the target data set are all face images that meet the quality requirements, and the accuracy is improved.

Based on the above embodiment, Fig. 3 is a schematic flow diagram of another data processing method provided in the embodiment of the present application. As shown in Fig. 3, the method includes the following steps:

Step 301, acquiring a target data set of a face image of a target object.

Step 302, input the face images in the image pair into the recognition model for feature extraction, and obtain the image features of the face images in the image pair.

The recognition model is a neural network model, for example, a recognition model trained based on the face recognition algorithm arcface, or a recognition model trained based on the face recognition algorithm cosface, which is not limited in this embodiment.

As an implementation, the recognition model includes a backbone network backbone and a fully connected layer of the head network, wherein the backbone network is, for example, a lightweight network mobilenet, which is used to output image features of a face image, and the image features are, for example, 512-dimensional features. The fully connected layer classifies based on image features to obtain classification results. In the face recognition scenario, the classification result is to identify the identity information of the face included in the face image by identifying the face image. The identity information can be The name, ID card information, etc. of the object to which the face belongs.

Step 303, according to the image features of the face images in the image pair, determine the distance between the image features.

Step 304, according to the distance between the image features, determine the similarity between the face images in the image pair.

As an implementation manner, the distance can be converted into a corresponding similarity according to the mapping relationship between the distance between image features and the similarity.

Wherein, step 303 and step 304 may refer to the explanations in the foregoing embodiments, and the principles are the same, so details are not repeated here.

Step 305, sort the distances, and determine the target distance from the sorting based on the set selection ratio.

Step 306: Determine the similarity threshold according to the target distance and the set mapping relationship.

In the embodiment of the present application, the distances between the target face images in each target image pair in the target data set are sorted, for example, the distances are sorted in order from large to small, and the sorting results are obtained. According to the target data set The cleaning accuracy requires to determine the set selection ratio, such as 5%, 7% or 10%, etc., and then select the corresponding target distance from the distance sorting results according to the set selection ratio, and according to the target distance and the set mapping relationship, Determine the similarity corresponding to the target distance, and use the similarity as the similarity threshold.

As an example, 20 target image pairs are determined from the target data set, and the distances between the target face images in the 20 target image pairs are s1, s2, s3, s4...s19 and s20 respectively. Set the selection ratio to 20%, that is, select 4 distances from 20 distances, then sort the above 20 distances in descending order, and the top 4 distances are s4, s7, s15 and s11, Then s11 is determined as the target distance, so that according to the mapping relationship between the target distance and the similarity, the similarity threshold corresponding to the target distance is determined, which improves the accuracy of determining the similarity threshold.

In step 307, each similarity is compared with a similarity threshold, and a target image pair whose similarity is smaller than the similarity threshold is determined.

Step 308, according to the target face images included in the target image pair, determine the occurrence frequency of each target face image. Step 309, in response to the occurrence frequency of any target human face image being greater than the quantity threshold, deleting any target human face image from the target data set.

Wherein, step 307 to step 309 may refer to the explanations in the foregoing embodiments, and the principles are the same, so details are not repeated here.

Step 310, in response to the occurrence frequency of any target human face image being less than or equal to the number threshold, prohibiting any target human face image from being deleted from the target data set.

In the embodiment of the present application, the frequency of appearance of each target human face image is compared with the quantity threshold, and in response to the frequency of occurrence of any target human face image is less than or equal to the quantity threshold, indicating that the target human face image is a target that meets the quality requirements The face image of the target, thus prohibiting the deletion of the target face image from the target data set, retaining the face image that meets the quality requirements, improving the quality of the face image in the target data set, and improving the follow-up based on the target data set. Identify the training accuracy for further training of the model.

In the data processing method of the embodiment of the present application, by calculating the distance between the face image features in each image pair in the target data set, the similarity between the face images in the image pair is determined, and from the determined similarity In the poor target image pair, the frequency of the target face image in the target image pair is obtained through statistics. The higher the frequency, the greater the difference between the target face image and other face images in the target data set. The face image is more likely to belong to noise data or abnormal data. Further, based on the frequency of occurrence, the target data set is cleaned of noise data, so that the face image data in the target data set are all face images that meet the quality requirements, which improves the The precision of the target dataset.

Based on the above-mentioned embodiments, the embodiment of the present application provides a model training method. FIG. 4 is a schematic flowchart of another model training method provided in the embodiment of the present application. As shown in FIG. 4, the method includes the following steps:

Step 401, acquire a sample target data set.

Wherein, the sample target data set includes sample face images of the target object; the sample face images are labeled with the identity information of the included faces.

Among them, the identity information is used to uniquely indicate an object, that is, a user, including a name, an ID card, and the like.

Step 402, input any sample face image into the recognition model for classification, so as to determine the identity information obtained by predicting any sample face image.

In an implementation of the embodiment of the present application, any sample face image is input into the feature extraction layer of the classification model for feature extraction to obtain high-dimensional image features of any sample face image, such as 512-dimensional image features, and The image features of any sample face image are input into the fully connected layer of the classification model, and the fully connected layer classifies based on the image features to determine the identity information obtained by predicting any sample face image.

Wherein, the description of the structure of the recognition model in the foregoing embodiments is also applicable to this embodiment, and the principles are the same, so details are not repeated here.

Step 403, based on the difference between the predicted identity information and the marked identity information, train the recognition model.

In the embodiment of the present application, the recognition model is trained based on the difference between the predicted identity information and the marked identity information, that is, the parameters of the recognition model are adjusted based on the difference, and the recognition model after adjusting the parameters is used as a sample The target data set is trained, and multiple rounds of iterations are performed through multiple samples until the recognition model converges. As an implementation method, the trained recognition model can be tested using the sample target data set used for testing. If the difference between the predicted identity information and the marked identity information is less than the set threshold, the recognition model training is considered complete. .

For the training of the recognition model, optionally, the recognition model can be continuously adjusted by parameters, and after the training of the set number of rounds is performed, the test set containing multiple sample target data sets can be used for testing. If the face recognition recall If the rate is greater than the first threshold and the non-recognition rate of face recognition is less than the second threshold, it is considered that the recognition model training is completed. Wherein, the first threshold is, for example, 90%, and the second threshold is, for example, 1%.

It should be understood that if the sample target data set is a sample target data set that has not been processed by the data processing method of the embodiment of the present application, such as the target data set in step 101 in the above embodiment, it can be used to identify The model is initially trained, and the accuracy of the trained recognition model does not meet the final requirements, but it can be used to recognize the face images in the target data set to obtain the image features of the face images.

If the sample target data set is the sample target data set processed by the data processing method of the embodiment of the present application, the recognition model can be trained, so that the trained recognition model has higher precision and improves the training effect of the model.

In the model training method of the embodiment of the present application, by training the recognition model, the difference between the identity information of the face image recognized by the trained recognition model and the identity information of the marked face image is smaller than the set threshold, that is The accuracy requirement is met, so that the face image features output by the recognition model have higher accuracy, so that the effect of data cleaning on the target data set can be improved.

In order to implement the foregoing embodiments, the embodiments of the present application further provide a data processing device.

FIG. 5 is a schematic structural diagram of a data processing device provided by an embodiment of the present application.

As shown in Figure 5, the device may include:

The acquisition module 51 is configured to acquire a target data set of face images of a target object; wherein, the target data set includes at least one image pair, and the face images in the image pair are different.

The first determination module 52 is configured to determine a target image pair whose similarity is smaller than a similarity threshold according to the similarity between the face images in the at least one image pair.

The second determination module 53 is configured to determine the frequency of appearance of each of the target human face images according to the target human face images included in the target image pair.

The processing module 54 is configured to delete any one of the target human face images from the target data set in response to the occurrence frequency of any one of the target human face images being greater than a quantity threshold.

Further, in an implementation manner of the embodiment of the present application, the first determining module 52 is specifically used for:

For each image pair, determine the similarity between the face images in the image pair;

Each of the similarities is compared with a similarity threshold, and a target image pair whose similarity is smaller than the similarity threshold is determined.

In an implementation manner of the embodiment of the present application, the first determination module 52 is specifically used to:

The human face image in the image pair is input into the recognition model to perform feature extraction, and the image features of the human face image in the image pair are obtained;

According to the image features of the face images in the image pair, determine the distance between the image features;

According to the distance, the similarity between the face images in the image pair is determined.

In an implementation manner of the embodiment of the present application, the processing module 54 is also used to:

In response to the occurrence frequency of any of the target human face images being less than or equal to the number threshold, it is forbidden to delete any of the target human face images from the target data set.

In an implementation manner of the embodiment of the present application, the method further includes:

The identification module is used to obtain an initial data set of the face image of the target object; identify the face image in the initial data set; recognize the face image whose face size is smaller than the set threshold, or, Face images that do not contain faces are deleted from the initial data set to obtain the target data set.

In an implementation manner of the embodiment of the present application, the method further includes:

The third determining module is configured to rank each of the distances; determine the target distance from the ranking based on a set selection ratio; and determine the similarity threshold according to the mapping relationship between the target distance and the similarity.

It should be noted that the foregoing explanations of the method embodiment are also applicable to the device of this embodiment, and details are not repeated here.

In the data processing device of the embodiment of the present application, the target data set of the face image of the target object is acquired, wherein the target data set includes at least one image pair, and the face images in the image pair are different, according to the person in the at least one image pair The similarity between the face images, determine the target image pair whose similarity is less than the similarity threshold, determine the frequency of occurrence of each target face image according to the target face image included in the target image pair, and respond to the appearance of any target face image If the frequency is greater than the number threshold, any target face image will be deleted from the target dataset. Through the similarity between face images, the target image pair with poor similarity is determined. Within the scope of the target image pair, the frequency of occurrence of the target face image in the target image pair is further counted, and the target image is calculated based on the frequency of occurrence. The data set is cleaned, so that the face image data in the target data set are all face images that meet the quality requirements, which improves the accuracy.

In order to implement the foregoing embodiments, an embodiment of the present application further proposes a model training device.

FIG. 6 is a schematic structural diagram of a model training device provided by an embodiment of the present application.

As shown in Figure 6, the device may include:

The acquisition module 61 is configured to acquire a sample target data set; wherein, the sample target data set contains sample human face images of target objects; the sample human face images are labeled with the identity information of the included human faces.

The recognition module 62 is configured to input any of the sample human face images into a recognition model for classification, so as to determine the identity information obtained by predicting any of the sample human face images.

The training module 63 is configured to train the recognition model based on the difference between the predicted identity information and the marked identity information.

It should be noted that the foregoing explanations of the method embodiment are also applicable to the device of this embodiment, and details are not repeated here.

In the model training device of the embodiment of the present application, by training the recognition model, the difference between the identity information of the face image recognized by the trained recognition model and the identity information of the marked face image is smaller than the set threshold, that is The accuracy requirement is met, so that the face image features output by the recognition model have higher accuracy, so that the effect of data cleaning on the target data set can be improved.

In order to realize the above embodiments, the present application also proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and operable on the processor. When the processor executes the program, the aforementioned method is implemented. The method described in the examples.

In order to implement the above embodiments, the present application also proposes a non-transitory computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method described in the foregoing method embodiments is implemented.

In order to implement the above embodiments, the present application also proposes a computer program product, on which a computer program is stored, and when the computer program is executed by a processor, the method described in the foregoing method embodiments is implemented.

FIG. 7 is a block diagram of an electronic device provided by an embodiment of the present application. For example, the electronic device 800 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, and the like.

7, electronic device 800 may include one or more of the following components: processing component 802, memory 804, power component 806, multimedia component 808, audio component 810, input/output (I/O) interface 812, sensor component 814, and communication component 816 .

The processing component 802 generally controls the overall operations of the electronic device 800, such as those associated with display, telephone calls, data communications, camera operations, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the above method. Additionally, processing component 802 may include one or more modules that facilitate interaction between processing component 802 and other components. For example, processing component 802 may include a multimedia module to facilitate interaction between multimedia component 808 and processing component 802 .

The memory 804 is configured to store various types of data to support operations at the electronic device 800 . Examples of such data include instructions for any application or method operating on the electronic device 800, contact data, phonebook data, messages, pictures, videos, and the like. The memory 804 can be implemented by any type of volatile or non-volatile storage device or their combination, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), Magnetic Memory, Flash Memory, Magnetic or Optical Disk.

The power component 806 provides power to various components of the electronic device 800 . Power components 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for electronic device 800 .

The multimedia component 808 includes a screen providing an output interface between the electronic device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from a user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensor may not only sense a boundary of a touch or swipe action, but also detect duration and pressure associated with the touch or swipe action. In some embodiments, the multimedia component 808 includes a front camera and/or a rear camera. When the electronic device 800 is in an operation mode, such as a shooting mode or a video mode, the front camera and/or the rear camera can receive external multimedia data. Each front camera and rear camera can be a fixed optical lens system or have focal length and optical zoom capability.

The audio component 810 is configured to output and/or input audio signals. For example, the audio component 810 includes a microphone (MIC), which is configured to receive external audio signals when the electronic device 800 is in operation modes, such as call mode, recording mode and voice recognition mode. Received audio signals may be further stored in memory 804 or sent via communication component 816 . In some embodiments, the audio component 810 also includes a speaker for outputting audio signals.

The I/O interface 812 provides an interface between the processing component 802 and a peripheral interface module, which may be a keyboard, a click wheel, a button, and the like. These buttons may include, but are not limited to: a home button, volume buttons, start button, and lock button.

Sensor assembly 814 includes one or more sensors for providing status assessments of various aspects of electronic device 800 . For example, the sensor component 814 can detect the open/closed state of the electronic device 800, the relative positioning of components, such as the display and the keypad of the electronic device 800, the sensor component 814 can also detect the electronic device 800 or a Changes in position of components, presence or absence of user contact with electronic device 800 , electronic device 800 orientation or acceleration/deceleration and temperature changes in electronic device 800 . Sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects in the absence of any physical contact. Sensor assembly 814 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor component 814 may also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor or a temperature sensor.

The communication component 816 is configured to facilitate wired or wireless communication between the electronic device 800 and other devices. The electronic device 800 can access a wireless network based on communication standards, such as WiFi, 4G or 5G, or a combination thereof. In an exemplary embodiment, the communication component 816 receives broadcast signals or broadcast related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on Radio Frequency Identification (RFID) technology, Infrared Data Association (IrDA) technology, Ultra Wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

In an exemplary embodiment, electronic device 800 may be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable A programmable gate array (FPGA), controller, microcontroller, microprocessor or other electronic component implementation for performing the methods described above.

In an exemplary embodiment, there is also provided a non-transitory computer-readable storage medium including instructions, such as the memory 804 including instructions, which can be executed by the processor 820 of the electronic device 800 to complete the above method. For example, the non-transitory computer readable storage medium may be ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, optical data storage device, and the like.

In the description of this specification, descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific examples", or "some examples" mean that specific features described in connection with the embodiment or example , structure, material or characteristic is included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the described specific features, structures, materials or characteristics may be combined in any suitable manner in any one or more embodiments or examples. In addition, those skilled in the art can combine and combine different embodiments or examples and features of different embodiments or examples described in this specification without conflicting with each other.

In addition, the terms "first" and "second" are used for descriptive purposes only, and cannot be interpreted as indicating or implying relative importance or implicitly specifying the quantity of indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include at least one of these features. In the description of the present application, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

Any process or method descriptions in flowcharts or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing custom logical functions or steps of a process , and the scope of preferred embodiments of the present application includes additional implementations in which functions may be performed out of the order shown or discussed, including in substantially simultaneous fashion or in reverse order depending on the functions involved, which shall It should be understood by those skilled in the art to which the embodiments of the present application belong.

The logic and/or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium, For use with instruction execution systems, devices, or devices (such as computer-based systems, systems including processors, or other systems that can fetch instructions from instruction execution systems, devices, or devices and execute instructions), or in conjunction with these instruction execution systems, devices or equipment for use. For the purposes of this specification, a "computer-readable medium" may be any device that can contain, store, communicate, propagate or transmit a program for use in or in conjunction with an instruction execution system, device or device. More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection with one or more wires (electronic device), portable computer disk case (magnetic device), random access memory (RAM), Read Only Memory (ROM), Erasable and Editable Read Only Memory (EPROM or Flash Memory), Fiber Optic Devices, and Portable Compact Disc Read Only Memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program can be printed, as it may be possible, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or other suitable processing if necessary. The program is processed electronically and stored in computer memory.

It should be understood that each part of the present application may be realized by hardware, software, firmware or a combination thereof. In the embodiments described above, various steps or methods may be implemented by software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented by any one or a combination of the following techniques known in the art: a discrete Logic circuits, ASICs with suitable combinational logic gates, Programmable Gate Arrays (PGA), Field Programmable Gate Arrays (FPGA), etc.

Those of ordinary skill in the art can understand that all or part of the steps carried by the methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. During execution, one or a combination of the steps of the method embodiments is included.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing module, each unit may exist separately physically, or two or more units may be integrated into one module. The above-mentioned integrated modules can be implemented in the form of hardware or in the form of software function modules. If the integrated modules are implemented in the form of software function modules and sold or used as independent products, they can also be stored in a computer-readable storage medium.

The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, and the like. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limitations on the present application, and those skilled in the art can make the above-mentioned The embodiments are subject to changes, modifications, substitutions and variations.

Claims (11)

1. A data processing method, characterized in that, comprising: Obtain a target data set of a face image of a target object; wherein, the target data set includes at least one image pair, and the face images in the image pair are different; According to the similarity between the face images in the at least one image pair, determine a target image pair whose similarity is less than a similarity threshold; According to the target human face images included in the target image pair, determine the frequency of occurrence of each of the target human face images; In response to the occurrence frequency of any of the target human face images being greater than a quantity threshold, any of the target human face images is deleted from the target data set. 2. The method according to claim 1, wherein, according to the similarity between the face images in the at least one image pair, determining the target image pair whose similarity is less than the similarity threshold includes: For each image pair, determine the similarity between the face images in the image pair; Each of the similarities is compared with a similarity threshold, and a target image pair whose similarity is smaller than the similarity threshold is determined. 3. The method according to claim 2, wherein, for each of the image pairs, determining the similarity between the face images in the image pairs includes: The human face image in the image pair is input into the recognition model to perform feature extraction, and the image features of the human face image in the image pair are obtained; According to the image features of the face images in the image pair, determine the distance between the image features; According to the distance, the similarity between the face images in the image pair is determined. 4. The method of claim 3, further comprising: sorting each of said distances; determining the target distance from the ranking based on the set selection ratio; The similarity threshold is determined according to the target distance and the set mapping relationship. 5. The method according to any one of claims 1-4, wherein the method further comprises: In response to the occurrence frequency of any of the target human face images being less than or equal to the number threshold, it is forbidden to delete any of the target human face images from the target data set. 6. The method according to any one of claims 1-4, wherein the method further comprises: Obtain an initial dataset of face images of the target object; Recognizing the face images in the initial data set; Deleting the recognized face images whose size is smaller than the set threshold, or the face images that do not contain a face, from the initial data set to obtain the target data set. 7. A method for model training, characterized in that the method comprises: Obtain a sample target data set; wherein, the sample target data set contains a sample face image of the target object; the sample face image is marked with the identity information of the included face; Inputting any of the sample human face images into the recognition model for classification to determine the identity information obtained by predicting any of the sample human face images; The recognition model is trained based on the difference between the predicted identity information and the labeled identity information. 8. A data processing device, characterized in that it comprises: An acquisition module, configured to acquire a target data set of a face image of a target object; wherein, the target data set includes at least one image pair, and the face images in the image pair are different; A first determination module, configured to determine a target image pair whose similarity is less than a similarity threshold according to the similarity between the face images in the at least one image pair; The second determination module is used to determine the frequency of appearance of each target face image according to the target face image included in the target image pair; A processing module, configured to delete any of the target face images from the target data set in response to the occurrence frequency of any of the target face images being greater than a quantity threshold. 9. A model training device, characterized in that the device comprises: An acquisition module, configured to acquire a sample target data set; wherein, the sample target data set contains a sample face image of a target object; the sample face image is marked with the identity information of the included face; A recognition module, configured to input any of the sample face images into a recognition model for classification, so as to determine the identity information obtained by predicting any of the sample face images; A training module, configured to train the recognition model based on the difference between the predicted identity information and the labeled identity information. 10. An electronic device, characterized in that it comprises a memory, a processor, and a computer program stored on the memory and operable on the processor, and when the processor executes the program, it realizes the process described in claims 1-6. Any described method, or realize the method as described in claim 7. 11. A non-transitory computer-readable storage medium, on which a computer program is stored, characterized in that, when the computer program is executed by a processor, the method according to any one of claims 1-6 is realized, or A method as claimed in claim 7.