Image classification method and device, electronic equipment and computer readable storage medium

Abstract

The application provides an image classification method and device, electronic equipment and computer readable storage medium; the method comprises: obtaining an image to be classified and an original label of the image to be classified; performing image classification on the image to be classified through a first image model to obtain a first classification label; in response to the first classification label being different from the original label, determining a first confidence of the image to be classified under a different image category through the first image model, and determining a second confidence of the image to be classified under the different image category through a second image model; the first image model is a pre-trained image classification model, and the second image model is an image classification model to be trained; based on the first confidence and the second confidence, performing image classification on the image to be classified to obtain an image classification result. Through the application, the accuracy of image classification can be improved.

Description

Technical Field

[0001] This application relates to the field of deep learning technology, and in particular to an image classification method, apparatus, electronic device, and computer-readable storage medium. Background Technology

[0002] With the rapid development of artificial intelligence and machine learning technologies, image recognition and classification tasks have been widely applied in numerous fields, such as autonomous driving, medical image diagnosis, and security monitoring. In these applications, high-quality image datasets are crucial for training high-performance models. However, labeling errors are a common problem in image datasets, severely impacting model training effectiveness and final performance. For example, image dataset labels are typically created manually. Due to factors such as the annotator's skill level, fatigue, and subjective judgment, labeling errors are inevitable. Furthermore, image datasets often contain a large number of images, covering various complex scenes and objects. In some cases, even experienced annotators struggle to accurately distinguish certain categories, leading to labeling errors. Moreover, in practical applications, it is difficult to separate incorrectly labeled sample images from the image dataset. Summary of the Invention

[0003] This application provides an image classification method, apparatus, electronic device, and computer-readable storage medium, which can improve the accuracy of image classification.

[0004] The technical solution of this application embodiment is implemented as follows:

[0005] This application provides an image classification method, the method comprising: acquiring an image to be classified and an original label of the image to be classified; classifying the image to be classified using a first image model to obtain a first classification label; in response to the first classification label being different from the original label, determining a first confidence level of the image to be classified under different image categories using the first image model, and determining a second confidence level of the image to be classified under the different image categories using a second image model; wherein the first image model is a pre-trained image classification model, and the second image model is an image classification model to be trained; and classifying the image to be classified based on the first confidence level and the second confidence level to obtain an image classification result.

[0006] This application provides an image classification apparatus, comprising: a data acquisition module for acquiring an image to be classified and its original label; a first image classification module for classifying the image to be classified using a first image model to obtain a first classification label; a confidence determination module for determining a first confidence level of the image to be classified under different image categories using the first image model, and determining a second confidence level of the image to be classified under the different image categories using a second image model, in response to the first classification label being different from the original label; the first image model being a pre-trained image classification model, and the second image model being a training image classification model; and a second image classification module for classifying the image to be classified based on the first confidence level and the second confidence level to obtain an image classification result.

[0007] In the above scheme, the confidence determination module is further used to perform image transformation processing on the image to be classified to obtain at least one transformed image; to summarize the image to be classified and the at least one transformed image to obtain a first image set; and to determine the first confidence of the images in the first image set under the different image categories through the first image model.

[0008] In the above scheme, the confidence determination module is further configured to, for each image category, determine the predicted confidence of each image in the first image set under the image category using the first image model; calculate the average of the predicted confidence of all images in the first image set to obtain the average confidence; and determine the average confidence as the first confidence of the images in the first image set under the image category.

[0009] In the above scheme, the second image classification module is further configured to determine the maximum first confidence level of the image to be classified under different image categories; determine the maximum second confidence level of the image to be classified under different image categories; and perform image classification on the image to be classified based on the maximum first confidence level and the maximum second confidence level to obtain the image classification result.

[0010] In the above scheme, the second image classification module is further configured to, in response to the maximum first confidence score being less than a preset first confidence score threshold and the maximum second confidence score being less than a preset second confidence score threshold, classify the image to be classified into a first image set; the first image set includes images whose original labels match the image information; in response to the maximum first confidence score being greater than or equal to the preset first confidence score threshold and / or the maximum second confidence score being greater than or equal to the preset second confidence score threshold, obtain the first image features output by the first data processing layer of the first image model and the second image features output by the second data processing layer of the second image model; perform image classification on the image to be classified based on the first image features and the second image features to obtain the image classification result.

[0011] In the above scheme, the second image classification module is further configured to determine the feature similarity between the first image feature and the second image feature; in response to the feature similarity being greater than a preset similarity threshold, classify the image to be classified into the first image set; in response to the feature similarity being less than or equal to the preset similarity threshold, classify the image to be classified into the second image set; the second image set includes images whose original labels do not match the image information.

[0012] In the above scheme, the second image classification module is further used to obtain the current training round number of the second image model, the preset first hyperparameter and the preset second hyperparameter; determine the second confidence threshold at the current time based on the current training round number and the first hyperparameter; and determine the similarity threshold at the current time based on the current training round number and the second hyperparameter.

[0013] In the above scheme, the first image classification module is further configured to classify the image to be classified into a third image set in response to the first classification label being the same as the original label; the third image set includes images whose original labels match the image information.

[0014] This application provides an electronic device, which includes: a memory for storing computer-executable instructions or computer programs; and a processor for executing the computer-executable instructions or computer programs stored in the memory to implement the image classification method provided in this application.

[0015] This application provides a computer-readable storage medium storing a computer program or computer-executable instructions for implementing the image classification method provided in this application when executed by a processor.

[0016] This application provides a computer program product, including a computer program or computer executable instructions, which, when executed by a processor, implement the image classification method provided in this application.

[0017] The embodiments of this application have the following beneficial effects:

[0018] When classifying images, the process begins with a pre-trained image classification model. This first classifies the image to be classified, obtaining a first classification label. If the first classification label differs from the original label, a first confidence level and a second confidence level are determined for each image category. Based on these first and second confidence levels, further image classification is performed. In other words, by comparing the first classification label with the original label, a preliminary coarse classification is performed. Then, when the first classification label differs from the original label, the pre-trained and untrained image classification models determine the first and second confidence levels for each image category. Finally, based on the first and second confidence levels, the image is classified to obtain the final classification result. Thus, since the pre-trained image classification model can provide more reliable confidence information, when performing further image classification using the confidence levels determined by the two models, the first confidence level obtained by the pre-trained image classification model can be used as a reference index. Based on this reference index, a more accurate classification result can be determined for the image to be classified, thereby improving the accuracy of image classification. Attached Figure Description

[0019] Figure 1 This is an optional flowchart illustrating the image classification method provided in an embodiment of this application;

[0020] Figure 2 This is another optional flowchart illustrating the image classification method provided in the embodiments of this application;

[0021] Figure 3 This is a schematic diagram of the implementation process for determining the first confidence level provided in an embodiment of this application;

[0022] Figure 4 This is a schematic diagram illustrating the implementation process of image classification provided in an embodiment of this application;

[0023] Figure 5 This is a schematic diagram illustrating the principle of the image classification method provided in the embodiments of this application;

[0024] Figure 6 This is a structural block diagram of an image classification device provided in an embodiment of this application;

[0025] Figure 7 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0027] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0028] In the following description, the terms "first, second, third" are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first, second, third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0029] In this application embodiment, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0030] Unless otherwise defined, all technical and scientific terms used in the embodiments of this application have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in the embodiments of this application is for the purpose of describing the embodiments of this application only and is not intended to limit this application.

[0031] In the implementation of this application, the collection and processing of relevant data should strictly comply with the requirements of relevant laws and regulations, obtain the informed consent or separate consent of the personal information subject, and carry out subsequent data use and processing within the scope of laws and regulations and the authorization of the personal information subject.

[0032] Before providing a further detailed description of the embodiments of this application, the nouns and terms involved in the embodiments of this application will be explained, and the nouns and terms involved in the embodiments of this application shall be interpreted as follows.

[0033] 1) Responding to: used to indicate the conditions or states on which the operation is performed depends. When the conditions or states on which it depends are met, one or more operations can be performed in real time or with a set delay. Unless otherwise specified, there is no restriction on the order in which the multiple operations are performed.

[0034] 2) Human-computer interaction interface, which is used to provide human-computer interaction functions / interface to display image classification information.

[0035] For example, graphical user interfaces (GUIs) include augmented reality (AR) interfaces, virtual reality (VR) interfaces, voice user interfaces (VUIs), interactive projection interfaces (using projection technology to display information on a flat surface), eye-tracking interfaces (interfaces controlled by detecting the user's gaze), holographic interfaces (three-dimensional holograms formed by projecting images using holographic projection technology, allowing users to see stereoscopic images without wearing special glasses), multimodal interfaces (interfaces that combine multiple interaction methods, such as tactile, visual, and auditory interaction), and brain-machine interfaces (BMIs).

[0036] In image classification tasks, noise in the labels of sample data (such as mislabeling or inaccuracy) leads to a decline in the performance of image classification models. When fine-tuning image classification models on datasets with noisy labels, overfitting and poor performance are likely to occur. Noisy labels cause image classification models to learn incorrect information, thereby affecting the accuracy of image classification models in classifying new images and reducing generalization ability. Furthermore, related techniques struggle to accurately distinguish between noisy and clean samples.

[0037] To address at least one problem in related technologies, embodiments of this application provide an image classification method, apparatus, electronic device, and computer-readable storage medium. When classifying an image to be classified, firstly, a pre-trained image classification model is used to classify the image, obtaining a first classification label. When the first classification label differs from the original label of the image to be classified, a first confidence level and a second confidence level for the image to be classified under different image categories are further determined. Based on the first and second confidence levels, further image classification is performed on the image to be classified. In other words, by comparing whether the first classification label is the same as the original label, a preliminary coarse classification can be performed on the image to be classified. Then, when the first classification label differs from the original label, the first and second confidence levels for the image to be classified under different image categories can be determined using both the pre-trained and untrained image classification models. Finally, based on the first and second confidence levels, the image to be classified is classified to obtain the image classification result. Thus, since the pre-trained image classification model can provide more reliable confidence information, when performing further image classification using the confidence levels determined by the two models, the first confidence level obtained by the pre-trained image classification model can be used as a reference index. Based on this reference index, a more accurate classification result can be determined for the image to be classified, thereby improving the accuracy of image classification.

[0038] The image classification method provided in this application can be applied to electronic devices such as laptops, tablets, and desktop computers. This application does not impose any restrictions on the specific type of electronic device.

[0039] The image classification methods provided in the embodiments of this application can be executed by an electronic device, which can be a server or a terminal. That is, the image classification methods in the embodiments of this application can be executed by a server, by a terminal, or through interaction between a server and a terminal.

[0040] The image classification method provided in the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0041] Figure 1 This is an optional flowchart illustrating an image classification method provided in an embodiment of this application. This method can be applied to electronic devices. The following description uses an electronic device as a server as an example. Figure 1 As shown, the method includes the following steps S101 to S104:

[0042] Step S101: Obtain the image to be classified and its original label.

[0043] In this embodiment, it should be noted that the image to be classified refers to an image in a classification task that classifies the image as a whole, not an image in a classification task that classifies the objects contained in the image. The original label of the image to be classified refers to the label of the object in the image in a classification task that classifies the objects contained in the image. For example, in an image dataset, each sample image has a corresponding label, which is the original label. However, there may be cases where the label is inconsistent with the object contained in the image, i.e., a noisy sample image. This embodiment can classify whether each sample image is a noisy sample image.

[0044] There are various ways to acquire images to be classified: from public datasets and from real-world application scenarios, etc., and this application does not limit the methods used in this embodiment. For example, ImageNet is a public image classification dataset. ImageNet has over 1000 categories, each with a large number of images. The original labels of these images are the category names, such as "cat," "dog," and "car." For example, to classify objects in a scene, an image of the object is captured using an imaging sensor, and then the image is labeled using an automatic annotation tool. The annotation results can be used as the name of the image file when the image is saved, such as naming the image file "rose."

[0045] Step S102: Classify the image to be classified using the first image model to obtain the first classification label.

[0046] Here, the first image model is a pre-trained image classification model.

[0047] In this embodiment, the pre-trained image classification model can be a model that has been trained on a large number of sample images. It can be trained on a large-scale dataset (such as ImageNet). The pre-trained image classification model can recognize a variety of common image categories. The pre-trained image classification model can assign input images to predefined categories. For example, the pre-trained image classification model can classify images into categories such as "cat," "dog," and "car."

[0048] The image to be classified is input into a pre-trained image classification model. The pre-trained model determines the most likely category of the input image based on the features and patterns learned during training. The output is the category label predicted by the pre-trained image classification model, which is the first classification label.

[0049] For example, a pre-trained Residual Network (ResNet) model classifies an image named "dog.jpg" (e.g., a picture containing a dog). The "dog.jpg" image is input into the ResNet model. The ResNet model extracts features from the image and performs calculations using its internal neural network structure. Ultimately, the ResNet model outputs a probability distribution representing the likelihood of the image belonging to each category. For example, if the probability distribution output by the ResNet model shows that the category "dog" has the highest probability (e.g., 98%), then "dog" is the primary classification label.

[0050] In step S103, in response to the first classification label being different from the original label, a first confidence level of the image to be classified under different image categories is determined by a first image model, and a second confidence level of the image to be classified under different image categories is determined by a second image model.

[0051] Here, the second image model is the image classification model to be trained.

[0052] In this embodiment, when the first classification label differs from the original label, the image to be classified can be classified using a first image model to obtain a first confidence level of the image under different image categories. Furthermore, the image to be classified can be classified using a second image model to obtain a second confidence level of the image under different image categories. The first classification label is the category label predicted by the first image model. If the first classification label differs from the original label, it indicates that the original label of the image to be classified may be inaccurate, i.e., the original label does not match the image information.

[0053] For example, the original label of an image might be "dog". When the first image model classifies the image, it outputs a confidence score for each category. The confidence score represents the probability that the image model predicts the image to be classified as that category; it's typically a value between 0 and 1, and the category corresponding to the highest confidence score is the classification label. For instance, the first image model can classify the images as "cat", "dog", and "car". The confidence scores might be 0.5 for "cat", 0.4 for "dog", and 0.1 for "car", meaning the first image model predicts a 50% probability that the image is a "cat", a 40% probability that it's a "dog", and only a 10% probability that it's a "car".

[0054] Similarly, the second image model also classifies the image to be classified and outputs a classification label and a confidence score (i.e., the second confidence score). The second image model is the image classification model to be trained, so its prediction results may differ from those of the first image model. For example, the second image model can classify the three categories "cat," "dog," and "car," obtaining confidence scores of 0.65 for "cat," 0.25 for "dog," and 0.1 for "car."

[0055] Step S104: Based on the first confidence level and the second confidence level, perform image classification on the image to be classified to obtain the image classification result.

[0056] In this embodiment, the image to be classified can be classified based on a first confidence level and a second confidence level to obtain the image classification result. It should be noted that the image classification here refers to a classification task that classifies the image as a whole.

[0057] According to the embodiments of this application, when classifying an image to be classified, firstly, a pre-trained image classification model is used to classify the image to be classified, obtaining a first classification label for the image to be classified. When the first classification label is different from the original label of the image to be classified, a first confidence level and a second confidence level of the image to be classified under different image categories are further determined, and then further image classification is performed on the image to be classified based on the first confidence level and the second confidence level. That is, by comparing whether the first classification label is the same as the original label, a preliminary coarse classification can be performed on the image to be classified. Then, when the first classification label is different from the original label, the first confidence level and the second confidence level of the image to be classified under different image categories can be determined by the pre-trained image classification model and the image classification model to be trained, respectively. Finally, the image to be classified is classified according to the first confidence level and the second confidence level to obtain the image classification result. Thus, since the pre-trained image classification model can provide more reliable confidence information, when performing further image classification using the confidence levels determined by the two models, the first confidence level obtained by the pre-trained image classification model can be used as a reference index. Based on this reference index, a more accurate classification result can be determined for the image to be classified, thereby improving the accuracy of image classification.

[0058] Figure 2 This is another optional flowchart illustrating the image classification method provided in the embodiments of this application, such as... Figure 2 As shown, the method includes the following steps S201 to S210:

[0059] Step S201: The terminal receives the image classification operation input by the user.

[0060] User-input image classification operations include selection operations or input operations. The selection operation is used to select the image to be classified, or the input operation is used to input the image identifier of the image to be classified.

[0061] In step S202, the terminal encapsulates the image identifier of the image to be classified into the image classification request.

[0062] In this embodiment of the application, the image classification request is used to request the server to classify the image to be classified.

[0063] In step S203, the terminal sends the image classification request to the server.

[0064] In this embodiment of the application, the terminal sends an image classification request to the server to request the server to classify the image to be classified.

[0065] In step S204, the server responds to the image classification request by obtaining the image to be classified and its original label.

[0066] In step S205, the server performs image classification on the image to be classified using the first image model to obtain the first classification label.

[0067] It should be noted that steps S204 to S205 are the same as steps S101 to S102 above. The implementation details of steps S204 to S205 in this embodiment will not be repeated.

[0068] In step S206, the server responds to the fact that the first classification label is the same as the original label and assigns the image to be classified to the third image set.

[0069] Here, the third image set includes images whose original labels match the image information.

[0070] In this embodiment, when the first classification label is the same as the original label, it can be considered that the original label of the image to be classified matches the image information, and the image to be classified is assigned to the third image set. If the first classification label is the same as the original label, it indicates that the original label of the image to be classified is accurate, that is, the original label matches the image information.

[0071] For example, using the first image model for classification, the first image model predicts the image as "dog". The first category label is "dog". The first category label "dog" is the same as the original label "dog". This indicates that the original label of the image matches the image content, and this image can be assigned to the third image set.

[0072] By step S206, by verifying whether the first classification label is the same as the original label, it can be verified whether the label of the image matches the image content. If the label of the image matches the image content, the image with the matching label can be classified into the third image set. The third image set can be used for subsequent high-quality dataset construction, model training or validation tasks, which helps to improve the quality of the dataset and the performance of the model.

[0073] In step S207, in response to the fact that the first classification label is different from the original label, the server determines the first confidence level of the image to be classified under different image categories using the first image model, and determines the second confidence level of the image to be classified under different image categories using the second image model.

[0074] Here, the first image model is a pre-trained image classification model, and the second image model is an image classification model to be trained.

[0075] In some embodiments, see Figure 3 , Figure 3 The step S207, "determining the first confidence level of the image to be classified under different image categories using the first image model," can be achieved through the following steps S2071 to S2073:

[0076] Step S2071: Perform image transformation processing on the image to be classified to obtain at least one transformed image.

[0077] In this embodiment, image transformation refers to performing a series of processing operations on an image to generate a new image. Image transformation methods can include: geometric transformations (such as rotation, scaling, translation, flipping, etc.), color transformations (such as adjusting brightness, contrast, saturation, etc.), noise addition (such as Gaussian noise, salt-and-pepper noise, etc.), and filtering operations (such as Gaussian blur, sharpening, etc.). Image transformation can increase the diversity of images.

[0078] For example, if the image to be classified is an image of a dog (the original image), the following transformations can be performed: rotate the image by 45 degrees, reduce the image brightness, and add salt and pepper noise to the image. Through these transformations, three transformed images can be obtained: the rotated image "dog01.jpg", the image with reduced brightness "dog02.jpg", and the image with added noise "dog03.jpg".

[0079] Step S2072: Summarize the images to be classified and at least one transformed image to obtain a first image set.

[0080] In this embodiment, the original image to be classified and all transformed images are combined to form an image set. This set can be called the first image set. The first image set contains the original image and all transformed images.

[0081] For example, the first image set may include “dog.jpg”, “dog01.jpg”, “dog02.jpg”, and “dog03.jpg”.

[0082] Step S2073: Determine the first confidence level of the images in the first image set under different image categories using the first image model.

[0083] In this embodiment, the first image model is a pre-trained image classification model, such as ResNet or MobileNet. The first image model can classify the input image and output the confidence score for each category.

[0084] For example, the first image model can classify three categories: "cat," "dog," and "car." When classifying the image "dog.jpg," the first image model obtains confidence scores of 0.5 for "cat," 0.4 for "dog," and 0.1 for "car." When classifying the transformed image "dog01.jpg," the first image model obtains confidence scores of 0.6 for "cat," 0.3 for "dog," and 0.1 for "car." When classifying the transformed image "dog02.jpg," the first image model obtains confidence scores of 0.6 for "cat," 0.4 for "dog," and 0.0 for "car." When classifying the transformed image "dog03.jpg," the first image model obtains confidence scores of 0.5 for "cat," 0.4 for "dog," and 0.1 for "car." The confidence scores of all images can be summarized to form a confidence matrix or list. Subsequently, the maximum confidence level corresponding to each category can be used as the first confidence level of the image in the first image set under the image category; alternatively, the mean of the confidence levels corresponding to each category can be used as the first confidence level of the image in the first image set under the image category; or the weighted sum of the confidence levels corresponding to each category can be used as the first confidence level of the image in the first image set under the image category. This application embodiment does not limit this, and the method of determining the first confidence level of the image in the first image set under the image category can be selected according to the actual situation.

[0085] In some embodiments, step S2073 can be implemented by performing the following process: First, for each image category, the prediction confidence of each image in the first image set under the image category is determined by the first image model; then, the average prediction confidence of all images in the first image set is calculated to obtain the average confidence; finally, the average confidence is determined as the first confidence of the image in the first image set under the image category.

[0086] In this embodiment, firstly, for each image category, the first image model is used to classify each image in the first image set, thereby obtaining the prediction confidence score of each image under each image category. For example, the prediction confidence scores of the original image "dog.jpg" are 0.5 for "cat", 0.4 for "dog", and 0.1 for "car". The prediction confidence scores of transformed image 1 "dog01.jpg" are 0.6 for "cat", 0.3 for "dog", and 0.1 for "car". The prediction confidence scores of transformed image 2 "dog02.jpg" are 0.6 for "cat", 0.4 for "dog", and 0.0 for "car". The prediction confidence scores of transformed image 3 "dog03.jpg" are 0.5 for "cat", 0.4 for "dog", and 0.1 for "car". Then, the average prediction confidence scores of all images in the first image set are calculated to obtain the average confidence score. The prediction confidence scores of each category can be collected. For the "cat" category: the prediction confidence scores for "dog.jpg" are 0.5, "dog01.jpg" are 0.6, "dog02.jpg" are 0.6, and "dog03.jpg" are 0.5. For the "dog" category: the prediction confidence scores for "dog.jpg" are 0.4, "dog01.jpg" are 0.3, "dog02.jpg" are 0.4, and "dog03.jpg" are 0.4. For the "Car" category: the prediction confidence scores for "dog.jpg" are 0.1, "dog01.jpg" are 0.1, "dog02.jpg" are 0.0, and "dog03.jpg" are 0.0. Next, the mean confidence score for each category is calculated. For the "Cat" category: mean confidence score = (0.5 + 0.6 + 0.6 + 0.5) / 4 = 0.55. For the "Dog" category: the mean confidence score is... The mean confidence score is (0.4 + 0.3 + 0.4 + 0.4) / 4 = 0.375. For the "Car" category, the mean confidence score is (0.1 + 0.1 + 0.0 + 0.1) / 4 = 0.075. Finally, the first confidence score for the images in the first image set under the "Cat" category is 0.55. The first confidence score for the images in the first image set under the "Dog" category is 0.375. The first confidence score for the images in the first image set under the "Car" category is 0.075.

[0087] Through the above processing, transformed images can be introduced, increasing the diversity of image data. This allows the first image model to evaluate the classification results of images from multiple perspectives, thereby improving the accuracy of classification. Furthermore, by comprehensively considering the classification results of the original image and the transformed image, random errors are reduced, enhancing the adaptability and reliability of the first image model in the face of different scenarios and improving the classification accuracy of the first image model.

[0088] Through steps S2071 to S2073, image transformation processing can be performed on the images to be classified to generate a diverse set of images (the first image set), which increases the diversity of image data. The pre-trained first image model is used to classify the images, calculate the confidence level under each category, and then determine the final first confidence level by using the confidence level information of the original image and the transformed image, thereby improving the reliability of the classification results of the first image model.

[0089] In step S208, the server performs image classification on the image to be classified based on the first confidence level and the second confidence level, and obtains the image classification result.

[0090] In some embodiments, see Figure 4 , Figure 4 Step S208 can be achieved through the following steps S2081 to S2083:

[0091] Step S2081: Determine the maximum first confidence score of the image to be classified under different image categories.

[0092] For example, the prediction confidence scores for each category are as follows: For the "cat" category: "dog.jpg" has a prediction confidence score of 0.5, "dog01.jpg" has a prediction confidence score of 0.6, "dog02.jpg" has a prediction confidence score of 0.6, and "dog03.jpg" has a prediction confidence score of 0.5. For the "dog" category: "dog.jpg" has a prediction confidence score of 0.4, "dog01.jpg" has a prediction confidence score of 0.3, "dog02.jpg" has a prediction confidence score of 0.4, and "dog03.jpg" has a prediction confidence score of 0.4. For the "car" category: "dog.jpg" has a prediction confidence score of 0.1, "dog01.jpg" has a prediction confidence score of 0.1, "dog02.jpg" has a prediction confidence score of 0.0, and "dog03.jpg" has a prediction confidence score of 0.0. The maximum first confidence score is 0.6.

[0093] Step S2082: Determine the maximum second confidence score of the image to be classified under different image categories.

[0094] For example, the second confidence level can be 0.65 for "cat", 0.25 for "dog", and 0.1 for "car". The maximum second confidence level is 0.65.

[0095] Step S2083: Based on the maximum first confidence and the maximum second confidence, perform image classification on the image to be classified to obtain the image classification result.

[0096] In some embodiments, step S2083 can be implemented by performing the following processing: First, in response to the maximum first confidence score being less than a preset first confidence score threshold and the maximum second confidence score being less than a preset second confidence score threshold, the image to be classified is assigned to a first image set; the first image set includes images whose original labels match the image information; second, in response to the maximum first confidence score being greater than or equal to the preset first confidence score threshold and / or the maximum second confidence score being greater than or equal to the preset second confidence score threshold, the first image feature output by the first data processing layer of the first image model and the second image feature output by the second data processing layer of the second image model are obtained; finally, the image to be classified is classified based on the first image feature and the second image feature to obtain the image classification result.

[0097] In this embodiment, a first confidence threshold can be preset. The first confidence threshold can be set according to actual conditions, and this embodiment does not limit it. Since the first image model is a pre-trained image classification model, a fixed first confidence threshold can be set to evaluate the accuracy of the first image model's classification. If the maximum first confidence is greater than the first confidence threshold, the accuracy of the first image model's classification can be considered high. If the maximum first confidence is less than or equal to the first confidence threshold, the accuracy of the first image model's classification can be considered low.

[0098] A second confidence threshold can be preset, which can be a fixed value or a variable value. When the second confidence threshold is a variable value, the current training epoch of the second image model and the preset first hyperparameter can be obtained first; then, based on the current training epoch and the first hyperparameter, the second confidence threshold at the current time can be determined. For example, the first hyperparameter may include an initial threshold and a decay rate. The calculation formula for the second confidence threshold can be constructed using the current training epoch, the initial threshold, and the decay rate. As the training epoch changes, the second confidence threshold will also change accordingly, thus determining the second confidence threshold corresponding to the current epoch. By dynamically adjusting the confidence threshold, the filtering process of classification results can be controlled more flexibly, thereby improving the accuracy and reliability of the classification results in subsequent applications.

[0099] When the maximum first confidence score is less than a preset first confidence threshold, and the maximum second confidence score is less than a preset second confidence threshold, the image to be classified is considered semantically complex, and the first image model cannot accurately classify the objects in the image. Therefore, the image to be classified can be considered an image whose original label matches the image information, and it is assigned to the first image set. For example, the maximum first confidence score is 0.6, the maximum second confidence score is 0.65, and both the first and second confidence thresholds are 0.8. The maximum first confidence score of 0.6 is less than the first confidence threshold of 0.8, and the maximum second confidence score of 0.65 is less than the second confidence threshold of 0.8, so the image "dog.jpg" can be assigned to the first image set.

[0100] When the maximum first confidence level is greater than or equal to a preset first confidence threshold, and / or the maximum second confidence level is greater than or equal to a preset second confidence threshold, the first image feature output by the first data processing layer of the first image model and the second image feature output by the second data processing layer of the second image model can be obtained. For example, the maximum first confidence level is 0.6, and the maximum second confidence level is 0.65. Both the first confidence threshold and the second confidence threshold are 0.6. The maximum first confidence level of 0.6 equals the first confidence threshold of 0.6, and the maximum second confidence level of 0.65 is greater than the second confidence threshold of 0.6, thus the first image feature output by the first data processing layer of the first image model and the second image feature output by the second data processing layer of the second image model can be obtained. The first data processing layer can be a layer in the first image model, and can be a convolutional layer or a feature extraction layer. The second data processing layer can be a layer in the second image model, and can be a convolutional layer or a feature extraction layer. The first image feature is the feature output by the first data processing layer of the first image model. The second image feature is the feature output by the second data processing layer of the second image model.

[0101] After obtaining the first image features and the second image features, image classification can be performed on the image to be classified based on the first image features and the second image features to obtain the image classification result. First, the feature similarity between the first image features and the second image features can be determined; then, in response to the feature similarity being greater than a preset similarity threshold, the image to be classified is assigned to the first image set; or, in response to the feature similarity being less than or equal to the preset similarity threshold, the image to be classified is assigned to the second image set; the second image set includes images whose original labels do not match the image information.

[0102] In this embodiment, a similarity threshold can be preset. The similarity threshold can be a fixed value or a variable value. When the similarity threshold is a variable value, the current training epoch of the second image model and the preset second hyperparameter can be obtained. Based on the current training epoch and the second hyperparameter, the similarity threshold at the current moment is determined. For example, the second hyperparameter may include an initial threshold and a decay rate. A formula for calculating the similarity threshold can be constructed using the current training epoch, the initial threshold, and the decay rate. As the training epoch changes, the similarity threshold also changes accordingly, thereby determining the similarity threshold corresponding to the current epoch. By dynamically adjusting the similarity threshold, the filtering process of classification results can be controlled more flexibly, thereby improving the accuracy and reliability of the classification results in subsequent applications.

[0103] There are several ways to determine the feature similarity between the first image feature and the second image feature: calculating the cosine similarity between the first image feature and the second image feature as the feature similarity, calculating the Euclidean distance between the first image feature and the second image feature as the feature similarity, calculating the Manhattan distance between the first image feature and the second image feature as the feature similarity, and calculating the Hamming distance between the first image feature and the second image feature as the feature similarity, etc. The choice can be made according to the actual situation, and the embodiments of this application do not limit this.

[0104] When the feature similarity is greater than a preset similarity threshold, the image to be classified can be assigned to the first image set. Alternatively, when the feature similarity is less than or equal to the preset similarity threshold, the image to be classified can be assigned to the second image set; the second image set includes images whose original labels do not match the image information.

[0105] Through the above processing, two confidence thresholds can be set to filter and process the image classification results. When the maximum confidence of both models is lower than their respective confidence thresholds, the original label of the image is considered to match the content, and the image is assigned to the first image set. When at least one confidence score is higher than the threshold, the features of the two models are further extracted and the similarity is calculated. When the similarity is greater than the similarity threshold, the original label of the image is considered to match the content, and the image is assigned to the first image set. Alternatively, when the similarity is less than or equal to the similarity threshold, the original label of the image is considered not to match the content, and the image is assigned to the second image set. Based on the multi-condition judgment classification method, fine-grained image classification can be achieved, effectively improving the accuracy and reliability of image classification. Furthermore, by dynamically adjusting the similarity threshold and confidence threshold, the filtering process of classification results can be controlled more flexibly, thereby improving the accuracy and reliability of the classification results in subsequent applications.

[0106] Through steps S2081 to S2083, the maximum confidence scores predicted by the two models for the image to be classified can be determined first. Then, a first confidence threshold and a second confidence threshold are set to filter and process the image classification results. When the maximum confidence scores of both models are lower than their respective confidence thresholds, the original label of the image can be considered to match the content, and the image is classified into the first image set. When at least one confidence score is equal to or higher than the threshold, the features of the two models are further extracted and further classification is performed. By dynamically adjusting the confidence threshold and similarity threshold, images whose labels may not match the image information can be effectively filtered out. At the same time, calculating feature similarity improves the accuracy and reliability of classification.

[0107] In step S209, the server sends the image classification results to the terminal.

[0108] In step S210, the terminal displays the image classification results.

[0109] The following will describe an exemplary application of the embodiments of this application in a real-world application scenario.

[0110] This application proposes an image classification method that can be based on image noise processing techniques and sample separation strategies guided by external factors (large image model, or a first image model) to solve the problem of noise labels affecting the performance of image classification models in image classification tasks, thereby improving the accuracy and robustness of image classification models.

[0111] See Figure 5 , Figure 5 This is a schematic diagram illustrating the principle of the image classification method provided in this application embodiment. First, a suitable large-scale image model (i.e., the first image model) is selected. A deep learning model pre-trained on a large-scale image dataset can be chosen based on the specific image task and the characteristics of the image dataset. The deep learning model can have image feature extraction capabilities, providing a foundation for subsequent sample analysis and model fine-tuning.

[0112] After selecting a suitable large-scale image model, confidence scores and features are obtained. Images from image data 501 can be used as sample images (i.e., images to be classified) and input into the selected large-scale image model 502. The prediction probabilities of the large-scale image model for each sample image belonging to each category (i.e., different image categories) are obtained, and the prediction probabilities of each category are used together as the confidence score of the sample image. Simultaneously, image features output from the intermediate layer of the large-scale image model can be extracted, namely, the intermediate features 508 of the large-scale image model (i.e., the first image features). The intermediate features of the large-scale image model contain rich semantic and structural information of the sample images, which can be used for subsequent sample separation and learning strategy design.

[0113] After obtaining the confidence level and features, coarse-grained separation is performed. The predicted label 504 (i.e., the first classification label) of the large image model for the sample image is compared with the original label 503 (i.e., the original label) of the sample image. The sample images are then divided into an easy-to-clean set (a potentially clean set of samples, i.e., the third image set) and an inconsistent set. Based on the consistency between the category predicted by the large image model and the original label, the sample images are initially divided into two categories. For example, if the category predicted by the large image model is consistent with the original label, the sample image is considered a potentially clean sample (and can be classified into the easy-to-clean set); if it is inconsistent, it is considered a potentially noisy sample (and can be classified into the inconsistent set).

[0114] After coarse-grained separation, data augmentation is performed to enhance the confidence of large image model generation. To make the confidence of large image model generation more robust, M different augmentation methods (i.e., image transformation processing, such as geometric transformation, color transformation, Gaussian noise, and cropping) can be applied to each input image while keeping the semantics of the augmented samples unchanged, resulting in M ​​augmented samples (i.e., at least one transformed image).

[0115] By using a large image model to predict M types of augmented samples, M confidence levels can be obtained. (i.e., predict confidence level), and then use the confidence level Aggregate into a comprehensive confidence level 506 (i.e., the first confidence level), which means that the overall confidence level is calculated separately for each category, as detailed in formula (1):

[0116]

[0117] in, The confidence level obtained for large image models M represents the confidence level generated by the large image model for each augmented sample, where M is the number of augmented samples.

[0118] Figure 5 The confidence level p505 (i.e., the second confidence level) is the confidence level obtained by the training model in predicting the sample image.

[0119] After data augmentation, fine-grained separation is performed. For the resulting potential noisy sample set (inconsistency set), the dynamic information of the trained image model during training (such as changes in prediction probability) and other feature information of the sample images (such as texture, color distribution, etc.) are further utilized to design more refined separation rules, dividing the sample images into a hard-clean set (which may be clean samples that were mistakenly identified as noise, i.e., the first image set) or a true noise set (i.e., the second image set), thereby more accurately identifying different types of samples.

[0120] After fine-grained separation, a threshold can be set. For the confidence level generated by large image models, a fixed confidence threshold can be used. (i.e., the first confidence threshold) is used for filtering. As for the confidence generated by the training model, since the training model will change during the fine-tuning process, an adaptive confidence threshold τ(t) (i.e., the second confidence threshold) can be used. This threshold increases as the fine-tuning progresses, but it will not be too high. The formula for calculating the adaptive confidence threshold τ(t) is given in formula (2):

[0121] τ(t)=τ-exp(-ωt) (2)

[0122] Where τ and ω are hyperparameters controlling the threshold, which can be set to τ = 0.6 and ω = 0.5, and t represents the current number of fine-tuning rounds.

[0123] In addition, a feature similarity threshold Φ(t) (i.e., similarity threshold) can be set. The formula for calculating the feature similarity threshold Φ(t) is given in formula (3):

[0124] Φ(t)=Φ-exp(-ωt) (3)

[0125] Where Φ and ω are hyperparameters controlling the threshold, and t represents the current fine-tuning round number.

[0126] Using a fixed confidence threshold After filtering using the first confidence threshold (i.e., the hard-wash set and the true noise set), the set can be selected. Based on the two set thresholds, the set that meets the requirements is selected. Images with max(p) < τ(t) are considered hard-washed. These images show low confidence in both the large-scale image model and the training model's predictions, possibly because they are semantically difficult to classify, yet their original labels are correct. Furthermore, the similarity between intermediate feature 507 (the second image feature) of the training model and intermediate feature 508 of the large-scale image model is calculated. Images with a similarity greater than the similarity threshold Φ(t) are considered hard-washed. Images with a similarity less than or equal to the threshold Φ(t) are classified as true noise. These images show high confidence in both the large-scale image model and the training model's predictions, but are considered noise samples due to inconsistent predictions in coarse-grained separation.

[0127] This application embodiment can filter out sample data with noisy labels (such as incorrect or inaccurate labeling) from the training dataset, and then correct the label information of the sample data. The corrected sample data can then be used to retrain the image classification model to be trained, thereby improving the performance of the image classification model.

[0128] Based on the image classification method described in the above embodiments Figure 6 The diagram shows a structural block diagram of an image classification device provided in an embodiment of this application. The image classification device 100 can be a device in an electronic device (e.g., a server). The image classification device can be implemented in software, which can be software in the form of programs and plug-ins, including the following software modules: data acquisition module 101, first image classification module 102, confidence determination module 103, and second image classification module 104. These modules are logically related, and therefore can be arbitrarily combined or further split according to the functions they implement.

[0129] The system includes a data acquisition module 101 for acquiring an image to be classified and its original label; a first image classification module 102 for classifying the image to be classified using a first image model to obtain a first classification label; a confidence determination module 103 for determining a first confidence level of the image to be classified under different image categories using the first image model in response to the first classification label being different from the original label, and for determining a second confidence level of the image to be classified under the different image categories using a second image model; the first image model is a pre-trained image classification model, and the second image model is an image classification model to be trained; and a second image classification module 104 for classifying the image to be classified based on the first confidence level and the second confidence level to obtain an image classification result.

[0130] In some embodiments, the confidence determination module 103 is further configured to perform image transformation processing on the image to be classified to obtain at least one transformed image; summarize the image to be classified and the at least one transformed image to obtain a first image set; and determine the first confidence level of the images in the first image set under the different image categories using the first image model.

[0131] In some embodiments, the confidence determination module 103 is further configured to, for each image category, determine the predicted confidence of each image in the first image set under the image category using the first image model; calculate the average of the predicted confidence of all images in the first image set to obtain the average confidence; and determine the average confidence as the first confidence of the images in the first image set under the image category.

[0132] In some embodiments, the second image classification module 104 is further configured to determine the maximum first confidence level of the image to be classified under different image categories; determine the maximum second confidence level of the image to be classified under different image categories; and perform image classification on the image to be classified based on the maximum first confidence level and the maximum second confidence level to obtain an image classification result.

[0133] In some embodiments, the second image classification module 104 is further configured to, in response to the maximum first confidence score being less than a preset first confidence score threshold and the maximum second confidence score being less than a preset second confidence score threshold, classify the image to be classified into a first image set; the first image set includes images whose original labels match the image information; in response to the maximum first confidence score being greater than or equal to the preset first confidence score threshold and / or the maximum second confidence score being greater than or equal to the preset second confidence score threshold, obtain the first image feature output by the first data processing layer of the first image model and the second image feature output by the second data processing layer of the second image model; perform image classification on the image to be classified based on the first image feature and the second image feature to obtain an image classification result.

[0134] In some embodiments, the second image classification module 104 is further configured to determine the feature similarity between the first image feature and the second image feature; in response to the feature similarity being greater than a preset similarity threshold, classify the image to be classified into the first image set; in response to the feature similarity being less than or equal to the preset similarity threshold, classify the image to be classified into the second image set; the second image set includes images whose original labels do not match the image information.

[0135] In some embodiments, the second image classification module 104 is further configured to obtain the current training epoch of the second image model, a preset first hyperparameter, and a preset second hyperparameter; determine the second confidence threshold at the current time based on the current training epoch and the first hyperparameter; and determine the similarity threshold at the current time based on the current training epoch and the second hyperparameter.

[0136] In some embodiments, the first image classification module 102 is further configured to classify the image to be classified into a third image set in response to the first classification label being the same as the original label; the third image set includes images whose original labels match the image information.

[0137] It should be noted that the description of the apparatus in this application embodiment is similar to the description of the method embodiment described above, and has similar beneficial effects as the method embodiment; therefore, it will not be repeated. For technical details not disclosed in this apparatus embodiment, please refer to the description of the method embodiment of this application for understanding.

[0138] This application provides an electronic device. Figure 7 This is a schematic diagram of the structure of the electronic device provided in an embodiment of this application. For example... Figure 7 As shown, the electronic device 130 includes: at least one processor 131 ( Figure 7 (Only one is shown in the image classification method embodiment above), memory 132, and computer-executable instructions 133 stored in memory 132 and executable on at least one processor 131, wherein processor 131 executes computer-executable instructions 133 to implement the steps in any of the above-described image classification method embodiments.

[0139] The electronic device may include, but is not limited to, a processor 131 and a memory 132. Those skilled in the art will understand that... Figure 7 This is merely an example of electronic device 130 and does not constitute a limitation on electronic device 130. It may include more or fewer components than shown, or combine certain components, or different components, such as input / output devices, network access devices, etc.

[0140] Processor 131 can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0141] In some embodiments, memory 132 may be an internal storage unit of electronic device 130, such as a hard disk or memory of electronic device 130. In other embodiments, memory 132 may be an external storage device of electronic device 130, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on electronic device 130. Furthermore, memory 132 may include both internal and external storage units of electronic device 130. Memory 132 is used to store operating system, application programs, bootloader, data, and other programs, such as program code of computer programs. Memory 132 may also be used to temporarily store data that has been output or will be output.

[0142] This application provides a computer program product comprising a computer program or computer-executable instructions stored in a computer-readable storage medium. The processor of an electronic device reads the computer-executable instructions from the computer-readable storage medium and executes the computer-executable instructions, causing the electronic device to perform the image classification method described in this application.

[0143] This application provides a computer-readable storage medium storing computer-executable instructions or a computer program. When the computer-executable instructions or the computer program are executed by a processor, the processor will execute the image classification method provided in this application. For example, ... Figure 1 The image classification method shown.

[0144] In some embodiments, the computer-readable storage medium may be a memory such as RAM, ROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or it may be a variety of devices including one or any combination of the above-mentioned memories.

[0145] In some embodiments, computer-executable instructions may take the form of programs, software, software modules, scripts, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as stand-alone programs or as modules, components, subroutines, or other units suitable for use in a computing environment.

[0146] As an example, computer-executable instructions may, but do not necessarily, correspond to files in a file system. They may be stored as part of a file that holds other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple co-located files (e.g., files that store one or more modules, subroutines, or code sections).

[0147] As an example, computer-executable instructions can be deployed to execute on a single electronic device, or on multiple electronic devices located at one location, or on multiple electronic devices distributed across multiple locations and interconnected via a communication network.

[0148] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, and improvements made within the spirit and scope of this application are included within the scope of protection of this application.

Claims

1. An image classification method, characterized in that, The method includes: Obtain the image to be classified and its original label; The image to be classified is classified using a first image model to obtain a first classification label; In response to the first classification label being different from the original label, a first confidence level of the image to be classified under different image categories is determined by the first image model, and a second confidence level of the image to be classified under the different image categories is determined by the second image model; the first image model is a pre-trained image classification model, and the second image model is an image classification model to be trained; Based on the first confidence level and the second confidence level, the image to be classified is classified to obtain the image classification result.

2. The method according to claim 1, characterized in that, The step of determining the first confidence level of the image to be classified under different image categories using the first image model includes: The image to be classified is subjected to image transformation processing to obtain at least one transformed image; The images to be classified and the at least one transformed image are summarized to obtain a first image set; Using the first image model, determine the first confidence level of the images in the first image set under the different image categories.

3. The method according to claim 2, characterized in that, The step of determining the first confidence level of images in the first image set under different image categories using the first image model includes: For each image category, the prediction confidence of each image in the first image set under the image category is determined using the first image model; The mean confidence score is obtained by averaging the prediction confidence scores of all images in the first image set. The mean confidence score is determined as the first confidence score of the images in the first image set under the image category.

4. The method according to claim 1, characterized in that, The step of classifying the image to be classified based on the first confidence level and the second confidence level to obtain the image classification result includes: Determine the maximum first confidence score of the image to be classified under the different image categories; Determine the maximum second confidence score of the image to be classified under the different image categories; Based on the maximum first confidence level and the maximum second confidence level, the image to be classified is classified to obtain the image classification result.

5. The method according to claim 4, characterized in that, The step of classifying the image to be classified based on the maximum first confidence score and the maximum second confidence score to obtain the image classification result includes: In response to the maximum first confidence score being less than a preset first confidence score threshold and the maximum second confidence score being less than a preset second confidence score threshold, the image to be classified is assigned to a first image set; the first image set includes images whose original labels match the image information; In response to the maximum first confidence level being greater than or equal to a preset first confidence level threshold, and / or the maximum second confidence level being greater than or equal to a preset second confidence level threshold, the first image feature output by the first data processing layer of the first image model and the second image feature output by the second data processing layer of the second image model are obtained; Based on the first image features and the second image features, the image to be classified is classified to obtain the image classification result.

6. The method according to claim 5, characterized in that, The step of classifying the image to be classified based on the first image features and the second image features to obtain the image classification result includes: Determine the feature similarity between the first image feature and the second image feature; In response to the feature similarity being greater than a preset similarity threshold, the image to be classified is assigned to the first image set; In response to the feature similarity being less than or equal to a preset similarity threshold, the image to be classified is assigned to a second image set; the second image set includes images whose original labels do not match the image information.

7. The method according to claim 6, characterized in that, The method further includes: Obtain the current training round number, preset first hyperparameter, and preset second hyperparameter of the second image model; Based on the current training round number and the first hyperparameter, determine the second confidence threshold at the current moment; Based on the current training round number and the second hyperparameter, the similarity threshold at the current time is determined.

8. The method according to any one of claims 1 to 7, characterized in that, The method further includes: In response to the first classification label being the same as the original label, the image to be classified is assigned to a third image set; the third image set includes images whose original labels match the image information.

9. An image classification device, characterized in that, The device includes: The data acquisition module is used to acquire the image to be classified and the original label of the image to be classified; The first image classification module is used to classify the image to be classified using a first image model to obtain a first classification label. A confidence determination module is configured to, in response to the first classification label being different from the original label, determine a first confidence level of the image to be classified under different image categories using a first image model, and determine a second confidence level of the image to be classified under the different image categories using a second image model; the first image model is a pre-trained image classification model, and the second image model is an image classification model to be trained; The second image classification module is used to classify the image to be classified based on the first confidence level and the second confidence level, and obtain the image classification result.

10. An electronic device, characterized in that, The electronic device includes: Memory is used to store executable instructions or computer programs. A processor, configured to execute computer-executable instructions or computer programs stored in the memory, implements the image classification method according to any one of claims 1 to 8.

11. A computer-readable storage medium storing computer-executable instructions or a computer program, characterized in that, When the computer-executable instructions or computer program are executed by a processor, the image classification method according to any one of claims 1 to 8 is implemented.

Images