Image quality partial order processing method and device, electronic equipment and storage medium

Abstract

Embodiments of the present disclosure disclose an image quality partial order processing method and device, electronic equipment and a storage medium. The method comprises: determining visual information and non-visual information of a to-be-processed image, the visual information comprising image features that can be obtained through visual elements in the to-be-processed image, the non-visual information comprising image features obtained through analysis of non-visual elements associated with the to-be-processed image, and performing an image quality determination operation by fusing the visual information and the non-visual information to obtain a predicted image quality of the to-be-processed image, and then performing an image partial order operation on the to-be-processed image according to the predicted image quality of the to-be-processed image. In the image quality determination, both information that can be easily and intuitively recognized, such as image content, and features that cannot be intuitively recognized are considered, ensuring that the image quality determination effectiveness can be achieved from the overall features of the picture, greatly improving the image partial order processing efficiency, and then automatically selecting high-quality images.

Description

Technical Field

[0001] This disclosure relates to the field of image processing technology, and in particular to an image quality partial order processing method, apparatus, electronic device, and storage medium. Background Technology

[0002] In many consumer scenarios that target a broad user base, such as category pages, the image content on category pages can be configured and sorted for display, allowing users to export and use it.

[0003] Currently, image quality is primarily characterized by manually assigning scores based on feedback. However, with the growing scale of the business system and the increasing number and categories of images, it's difficult to maintain regular updates through manual configuration. Furthermore, many images of low quality, with infrequent export usage, are displayed prominently and have high click-through rates, resulting in many high-quality images failing to receive adequate exposure and being missed for export. Therefore, analyzing images on similar category pages to optimize the display of high-quality images becomes crucial. Summary of the Invention

[0004] This disclosure provides an image quality partial order processing method, apparatus, electronic device, and storage medium to optimize the image quality evaluation process, automatically select high-quality images for optimized display, and greatly improve the efficiency and accuracy of image sorting.

[0005] In a first aspect, this disclosure provides an image quality partial order processing method, the method comprising:

[0006] Determine the visual and non-visual information of the image to be processed; the visual information includes image features that can be obtained from visual elements in the image to be processed, and the non-visual information includes image features obtained by parsing non-visual elements associated with the image to be processed;

[0007] The visual information and non-visual information are fused to perform an image quality determination operation, thereby obtaining the predicted image quality of the image to be processed.

[0008] A partial image ordering operation is performed on the image to be processed based on the predicted image quality.

[0009] Secondly, this disclosure also provides an image quality partial order processing apparatus, the apparatus comprising:

[0010] The information determination module is used to determine the visual information and non-visual information of the image to be processed; the visual information includes image features that can be obtained through visual elements in the image to be processed, and the non-visual information includes image features obtained by parsing non-visual elements associated with the image to be processed;

[0011] The image quality determination module is used to fuse the visual information and non-visual information to perform an image quality determination operation and obtain the predicted image quality of the image to be processed.

[0012] The image partial ordering module is used to perform image partial ordering operations on the image to be processed based on the predicted image quality of the image to be processed.

[0013] Thirdly, this disclosure also provides an electronic device, the electronic device comprising:

[0014] At least one processor; and

[0015] A memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the image quality partial order processing method described in any of the above embodiments.

[0017] Fourthly, this disclosure also provides a computer-readable medium storing computer instructions that cause a processor to execute the image quality partial order processing method described in any of the above embodiments.

[0018] The technical solution of this disclosure, when performing image partial order processing, determines the visual and non-visual information of the image to be processed. The visual information includes image features that can be obtained through visual elements in the image to be processed, and the non-visual information includes image features obtained by parsing non-visual elements associated with the image to be processed. The visual and non-visual information are fused to perform an image quality judgment operation to obtain the predicted image quality of the image to be processed. Then, based on the predicted image quality, the image partial order operation is performed on the image to be processed. By simultaneously introducing the visual and non-visual information of the image into the image quality judgment process, the image quality judgment considers both easily identifiable information such as the image content and features that cannot be directly identified, ensuring that the effectiveness of image quality judgment can be achieved from the overall features of the image. This greatly improves the efficiency of partial order processing between images and enables automatic screening of high-quality images.

[0019] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0020] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0021] Figure 1 A flowchart of an image quality partial order processing method provided in this embodiment of the disclosure;

[0022] Figure 2 A flowchart of another image quality partial order processing method provided in this disclosure embodiment;

[0023] Figure 3 A schematic diagram of the principle of the first image quality determination model provided in the embodiments of this disclosure;

[0024] Figure 4 A schematic diagram of the principle of the second image quality determination model provided in the embodiments of this disclosure;

[0025] Figure 5 A schematic diagram illustrating the applicable sorting of training sample images by image export click ratio, provided for embodiments of this disclosure;

[0026] Figure 6 A flowchart illustrating a testing method for an image quality assessment model provided in this embodiment of the disclosure;

[0027] Figure 7 This is a structural block diagram of an image quality partial order processing device provided in an embodiment of the present disclosure;

[0028] Figure 8 A structural block diagram of an electronic device for implementing the image quality partial order processing method of this disclosure. Detailed Implementation

[0029] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0030] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.

[0031] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below.

[0032] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0033] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0034] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0035] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0036] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the software or hardware, such as the electronic device, application, server, or storage medium performing the operations of this disclosed technical solution, based on the prompt message.

[0037] As an optional but non-limiting implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0038] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0039] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

[0040] In the following embodiments, each embodiment provides optional features and examples. The various features described in the embodiments can be combined to form multiple optional solutions. Each numbered embodiment should not be regarded as only one technical solution. Furthermore, unless otherwise specified, the embodiments and features in the embodiments of this disclosure can be combined with each other.

[0041] Figure 1 This is a flowchart illustrating an image quality partial order processing method provided in this embodiment. The technical solution of this embodiment is applicable to situations where partial order processing is performed between images to filter high-quality images, particularly for partial order processing of images displayed on a classification page. This method can be executed by an image quality partial order processing device, which can be implemented by software and / or hardware and is generally integrated into any electronic device with network communication capabilities, including but not limited to computers, personal digital assistants, etc. Figure 1 As shown, the image quality partial order processing method of this embodiment may include the following steps S110-S130:

[0042] S110. Determine the visual and non-visual information of the image to be processed; the visual information includes image features that can be obtained through visual elements in the image to be processed, and the non-visual information includes image features obtained by parsing the non-visual elements associated with the image to be processed.

[0043] The image to be processed can be an image that needs to be analyzed for partial order relationships with other images. For example, multiple images can be displayed in a sorted manner on a category page. Any image that needs to be displayed on the category page can be used as the image to be processed and analyzed for partial order relationships with other images on the category page.

[0044] The difference between visual and non-visual information lies in whether image features can be intuitively identified from the image to be processed through visual perception, thus enabling image differentiation. For the visual information of the image to be processed, visual information includes image features that can be obtained from the visual elements within the image. For example, visual information can include information that allows direct identification of image features from the content of the image to be processed, and visual information includes image features that can be intuitively obtained from the image to be processed through visual perception.

[0045] Non-visual information includes image features obtained through parsing non-visual elements associated with the image to be processed. For example, non-visual information may include features that cannot be directly identified from the content of the image. It also includes image features that cannot be directly obtained through visual perception but can be indirectly obtained through parsing the image. Visual information is highly correlated with human perception and is greatly influenced by subjectivity, which may lead to some bias when performing partial order analysis of images. Therefore, non-visual information from the image can be introduced when performing partial order processing on the image to be processed.

[0046] S120. The visual information and non-visual information are fused to perform an image quality determination operation to obtain the predicted image quality of the image to be processed.

[0047] As an optional but non-limiting implementation, visual information is used to measure the content attributes of an image. Visual information includes image features that can be intuitively obtained from the image through visual perception. Visual information can intuitively reflect the characteristics of the content in the image to be processed. For example, image content attributes can be described by at least one of the following: image color, image texture, image lines, and image content richness, etc., which are intuitively identifiable features.

[0048] As an optional but not limiting implementation, non-visual information is used to measure additional attributes of an image beyond its content. Non-visual information includes image features that cannot be directly obtained through visual perception but can be indirectly obtained through image analysis. For example, additional image attributes can be described by at least one of the following: image format, image color space type, image size, and image creation information (such as whether the publishing user is the image's creator). While non-visual information may not directly reflect the characteristics of the image's content, it can supplement visual information from a non-visual perspective. By measuring image size, calculating the shape of lines, etc., the diversity analysis of the image's content information can be performed and added to the visual information, thus addressing the shortcomings of visual information in partial order analysis and making the partial order analysis more comprehensive.

[0049] Optionally, the image format in this embodiment may include, but is not limited to, png, jpeg, gif, etc., the image color space type may include, but is not limited to, RGB, RGBA, etc., and the image size may include, but is not limited to, the width and height of the image, as shown in Table 1.

[0050] Table 1 Image-related attributes in non-visual information

[0051]

[0052] After obtaining the visual and non-visual information of the image to be processed, the visual and non-visual information can be fused together to participate in the image quality judgment operation. In this way, the image quality judgment process not only considers the impact of the image's visual information on image quality, such as the image's color, texture, lines, and content richness, but also uses some non-visual information of the image, such as image format and image color space type, to characterize the image's "height, width, and thickness" from another dimension. The useful information contained in the image fully supplements the deficiencies of the visual information, thereby improving the accuracy of image quality judgment.

[0053] S130. Perform image partial ordering operation on the image to be processed based on the predicted image quality of the image to be processed.

[0054] The predicted image quality of the image to be processed can, to some extent, reflect its importance when displayed on the classification page. This allows us to determine the partial order relationship between the image to be processed and other images based on the predicted image quality, thus guiding the ranking of the images to be processed on the classification page.

[0055] The higher the quality of the predicted image of the image to be processed, the earlier the image will be displayed on the classification page, and the better it will be displayed outwards; the lower the quality of the predicted image of the image to be processed, the later the image will be displayed on the classification page.

[0056] As an optional but non-limiting implementation, the predicted image quality of the image to be processed can be measured using, but is not limited to, the following metrics: Click Through Rate (CTR), Export Through Rate (ETR), and Export Click Rate (ECR). Specifically, the Click Through Rate is determined by the ratio of the number of times an image is clicked to the number of times it is viewed; the Export Through Rate is determined by the ratio of the number of times an image is exported to the number of times it is viewed; and the Export Click Rate is determined by the ratio of the number of times an image is exported to the number of times it is selected by clicking.

[0057] Image quality assessment primarily relies on manual scoring to construct training datasets, which is heavily influenced by subjective human perception and is difficult to annotate extensively. Therefore, defining image quality skew manually is challenging; only a small number of high-quality images can be selected from the entire dataset, making it impossible to manually annotate the overall image quality skew or to extract a large number of low-quality images as training negatives.

[0058] To address scenarios where manual annotation is difficult, a posterior metric for image consumption is used as a measure of model fit, directly representing image quality. This implies that users' image selection process implicitly includes a partial ranking of image quality. The posterior metric may include, but is not limited to, image click-to-display ratio, image export-to-display ratio, and image export-to-click ratio.

[0059] According to the technical solution of the present disclosure, when performing image partial order processing, by simultaneously introducing visual and non-visual information of the image into the image quality judgment process, the image quality judgment will consider both easily identifiable information such as the image content and features that cannot be directly identified, ensuring that the image quality judgment can be effective from the overall features of the image, greatly improving the efficiency of partial order processing between images, and thus enabling automatic screening of high-quality images.

[0060] Figure 2 This is a flowchart of another image quality partial order processing method provided by an embodiment of this disclosure. The technical solution of this embodiment further optimizes the process of fusing visual and non-visual information to perform image quality determination operations in the aforementioned embodiments, based on the above embodiments. This embodiment can be combined with various optional solutions in one or more of the above embodiments. Figure 2 As shown, the image quality partial order processing method of this embodiment may include the following steps S210-S240:

[0061] S210. Determine the visual and non-visual information of the image to be processed; visual and non-visual information are distinguished by whether features can be directly identified from the image.

[0062] S220. Input the visual information of the image to be processed into the first image quality judgment model, and obtain the visual features of the image to be processed by performing a visual feature extraction operation.

[0063] The first image quality assessment model is obtained by training the model using visual information from training sample images in the training set.

[0064] See Figure 3 In the first model training phase, the training set required for the first image quality judgment model is obtained, and the visual information of the training sample images in the training set is input into the first image quality judgment model to be trained. The feature vector of the visual information of the training sample images is calculated through the first image quality judgment model, and the prediction is directly fitted based on the calculated feature vector. The loss of the fitted prediction result is calculated using a preset loss function, and the model is trained using the calculated loss to obtain the trained second image quality judgment model.

[0065] Through the above model training, the first image quality judgment model can fully consider the impact of information such as image color, image texture, image lines, and the richness of image content on image quality judgment. In this way, the trained first image quality judgment model can be used to perform image quality judgment operation on the image to be processed. During the image quality judgment operation, visual features of the image to be processed that can be used for subsequent feature fitting are extracted.

[0066] As an optional but non-limiting implementation, the first image quality assessment model is obtained through iterative training based on a visual feature extraction network. Specifically, the first image quality assessment model employs a Convolutional Neural Network (CNN) framework, using a Swin-Transformer architecture as the visual feature extraction network, to enable the first image quality assessment model to extract visual features from images.

[0067] As an optional but non-limiting implementation, the first image quality assessment model can be trained using visual information from training sample images in the training set, with the image export click ratio as the image quality assessment metric.

[0068] For image click-to-show ratio, image export-to-show ratio, and image export-to-click ratio, since the image click-to-show ratio (CTR) and image export-to-show ratio (ETR) are greatly affected by the original display order of the image, while the image export-to-click ratio (ECR) is the probability of actually exporting the finished image after the user has already clicked on the image, which has eliminated the influence of the image display order and the user's judgment of image quality is more accurate, the image export-to-click ratio (ECR) is used as the fitting index for model training.

[0069] Optionally, the first image quality assessment model may use, but is not limited to, the root mean square error (RMSE) for loss calculation during model training. The formula for calculating the RMSE loss is as follows:

[0070]

[0071] in, This represents the quality of the pre-labeled image of the i-th training sample image. The image quality is represented by the predicted image quality for the i-th training sample image during model training, and n represents the number of training sample images participating in the training of the image quality judgment model. The image quality is measured by the image export click ratio.

[0072] S230. The non-visual features and visual features in the non-visual information are fused and input into the second image quality judgment model to obtain the predicted image quality of the image to be processed.

[0073] The second image quality assessment model is obtained by training the model using visual and non-visual information from the training sample images, based on the first image quality assessment model.

[0074] See Figure 4 In the second model training phase, the first image quality assessment model trained in the first model training phase can be used to infer from the training sample images in the training set, saving the visual features extracted from the training sample images. Simultaneously, non-visual features of the training sample images can also be obtained from their non-visual information. The non-visual and visual features of the training sample images are fused and input into the second image quality assessment model to be trained. The second image quality assessment model then fits and predicts the fused feature vector, calculates the loss on the fitted prediction result using a preset loss function, and trains the model using the calculated loss, resulting in the trained second image quality assessment model.

[0075] Through the training of the aforementioned model, the second image quality assessment model not only focuses on the richness of image content such as color, texture, and lines, but also incorporates raw non-visual information such as image width, height, and format. This allows the model to learn image features more comprehensively, judging not only the refinement of the image's "facial features" (i.e., local details) but also the overall image structure, enabling quality assessment from a holistic perspective. This allows the trained second image quality assessment model to be used to perform image quality assessment operations on the images being processed, effectively representing the overall image quality and improving the accuracy of image quality assessment. Furthermore, the reason for using the output of the first image quality assessment model for training is to better connect the processing procedures of the first and second image quality assessment models.

[0076] As an optional but non-limiting implementation, the second image quality assessment model is trained iteratively based at least on a gradient boosting framework network with a weak classifier. Specifically, the second image quality assessment model uses the LightGBM (Light Gradient Boosting Machine) framework as a gradient boosting framework network based on a weak classifier to perform image quality assessment operations after fusing visual and non-visual features.

[0077] The LightGBM framework described above is a fast, distributed, and high-performance gradient boosting framework based on the decision tree algorithm. LightGBM is based on iterative training of weak classifiers (decision trees), which can fully utilize discretized data features to achieve feature discrimination and data classification.

[0078] As an optional but non-limiting implementation, the second image quality assessment model is trained based on the first image quality assessment model by using the visual and non-visual information of the training sample images and the image export click ratio as the image quality assessment metric. See the foregoing for details.

[0079] Optionally, the second image quality assessment model may use, but is not limited to, root mean square error (RMSE) for loss calculation during model training.

[0080] As an optional but non-limiting implementation, the number of clicks on the training sample images is greater than a preset threshold, and the training sample images are evenly distributed across different image click ratios within a preset range after being sorted by their image export click ratios, rather than clustering within a small number of image click ratio ranges.

[0081] Based on the posterior metrics of business data, the ECR (Export Click Ratio) of each image is calculated as a partial metric for image quality, thus eliminating the need for manual scoring and automatically selecting and constructing data that can be used for model training, while also serving as a fitting metric for the algorithm model training.

[0082] See Figure 5 To ensure the reliability of training sample images, images with more than 50 clicks can be selected for the training set. For the sample images in the training set, the Image Export Click-Through Rate (ECR) is used as a metric for image quality. The ECR is calculated, and each image is sorted from highest to lowest. The results show that the ECR is evenly distributed and does not cluster in a single score range, thus effectively representing the image quality distribution. Therefore, by selecting images with sufficient posterior data and calculating their ECR, the selection of training set data for model training can be automated.

[0083] As an optional but non-limiting implementation, fusing non-visual features from non-visual information with the visual features and inputting the result into the second image quality assessment model may include steps A1-A3:

[0084] Step A1: Encode and map the non-visual features in the non-visual information to obtain discretized non-visual features.

[0085] Step A2: Fuse the visual features of the image to be processed with the discretized non-visual features to obtain multimodal features.

[0086] Step A3: Input the multimodal features into the second image quality assessment model to perform image quality assessment operation, and output the predicted image quality of the image to be processed.

[0087] The image table features are encoded and mapped separately to become discretized data features. The visual features of the image to be processed are calculated by the trained first image quality assessment model and combined with the discretized non-visual features to form a multimodal feature vector. The fused multimodal feature vector is then fed into the second image quality assessment model to perform image quality assessment and directly fit the predicted image quality.

[0088] As an optional but non-restrictive implementation, see [link to relevant documentation]. Figure 4 In the second model training stage, non-visual features in the non-visual information of the training sample images are encoded and mapped to obtain discretized non-visual features, which are then fused with visual features obtained by feature extraction of visual information from the training sample images through the first image quality judgment model. The fused multimodal feature vector is then input into the second image quality judgment model to be trained for fitting and prediction. A preset loss function is used to calculate the loss of the fitting and prediction results, and the calculated loss is used to train the model to obtain the trained second image quality judgment model.

[0089] S240. Perform image partial ordering operation on the image to be processed based on the predicted image quality of the image to be processed.

[0090] According to the technical solution of this disclosure, when performing partial order processing of images, by simultaneously introducing visual and non-visual information of the image into the image quality judgment process, a two-stage model is adopted. When judging image quality, it considers both easily identifiable information such as the image content and features that cannot be directly identified. This allows the model to learn image features more comprehensively, considering both the richness of visual features such as color, texture, and lines, and non-visual features such as image width, height, format, and creation information. This enables the model to judge whether the "facial features" of the image are exquisite, while also paying attention to the "height, width, and thickness" of the image, ensuring the effectiveness of image quality judgment based on the overall features of the image. This greatly improves the efficiency of partial order processing between images, and thus allows for the automatic selection of high-quality images.

[0091] Figure 6 This is a flowchart illustrating a method for testing an image quality judgment model in partial order processing of image quality, provided by an embodiment of this disclosure. The technical solution of this embodiment further optimizes the testing process of the first and second image quality judgment models trained in the aforementioned embodiments, based on the above embodiments. This embodiment can be combined with various optional solutions in one or more of the above embodiments. Figure 6 As shown, the image quality assessment model testing method of this embodiment may include the following steps S610-S630:

[0092] S610. Determine the test sample image and the reference image of the test sample image from the test set.

[0093] S620. Using the first image quality judgment model and the second image quality judgment model, the image quality of the test sample image and the reference image of the test sample image are judged to obtain the image quality of the test sample image and the predicted image quality of the reference image.

[0094] Once the first and second image quality assessment models are trained, the trained models can be tested using a test set. Considering that image quality calculation is for partial order processing, testing the model requires verifying whether the model's output satisfies the partial order relation. Therefore, a corresponding reference image needs to be configured for each test sample image in the test set.

[0095] The test sample image and its reference image are evaluated for image quality using a first image quality assessment model and a second image quality assessment model. This yields the predicted image quality of the test sample image and the reference image, allowing for the determination of the partial order relationship between the predicted image quality of the test sample image and the reference image. Furthermore, the pre-annotated image quality partial order relationship between the test sample image and the reference image can be statistically determined. This allows for the verification of the stability of the partial order during testing.

[0096] As an optional but non-limiting implementation, the difference between the image export click ratio of the test sample image and the reference image is greater than a preset multiple of the test standard deviation, which is determined based on the standard deviation of the image export click ratio of each test sample image in the test set.

[0097] Using each test sample image in the test set as a baseline, another image in the test set is randomly selected as a reference image. The difference between the test sample image and its reference image must be greater than the overall standard deviation of the ECR of the n test sample images in the test set. A difference of one standard deviation can be considered a threshold for significant difference, thus selecting a reference image for each test sample image. The preset multiplier can be one or more times.

[0098] S630. Statistically analyze the predicted image quality bias and the pre-labeled image quality bias between the test sample image and the reference image to determine whether to initiate the update of the first image quality judgment model and the second image quality judgment model.

[0099] As an optional but non-limiting implementation, statistical analysis of the predicted image quality skewness and the pre-labeled image quality skewness between the test sample image and the reference image may include steps B1-B3:

[0100] Step B1: If the predicted image quality bias between the test sample image and the reference image is consistent with the pre-labeled image quality bias, then the image quality judgment of the test sample image and the reference image is taken as a positive test example.

[0101] Step B2: If the predicted image quality bias between the test sample image and the reference image is inconsistent with the pre-labeled image quality bias, then the image quality judgment of the test sample image and the reference image shall be used as a negative test example.

[0102] Step B3: Count the number of positive test cases and the number of negative test cases, and calculate the ratio between the number of positive test cases and the number of negative test cases as the positive-to-negative ratio.

[0103] For each test sample image, image quality is assessed separately for both the test sample image and the reference image. The number of positive and negative test examples during this assessment is counted, and the ratio of positive to negative examples is calculated as the positive-to-negative order ratio. Experimental analysis shows that based solely on visual features, the positive-to-negative order ratio is 2.6. However, based on multimodal features fused from both visual and non-visual features, the ratio is 3.3 on the same test set. This demonstrates that non-visual features significantly improve the model's partial order performance.

[0104] The larger the ratio of forward to reverse order, the more obvious the partial order effect of the first and second image quality judgment models in determining image quality; the smaller the ratio of forward to reverse order, the less obvious the partial order effect of the first and second image quality judgment models in determining image quality.

[0105] If the ratio of forward to reverse order is determined to be less than a preset value, the training and update operation of the first image quality judgment model and the second image quality judgment model is initiated to continue optimizing and training the first image quality judgment model and the second image quality judgment model.

[0106] Figure 7 This is a structural block diagram of an image quality partial order processing device provided in an embodiment of this disclosure. The technical solution of this embodiment can be applied to the situation of partial order processing between images to filter high-quality images, especially the situation of partial order processing of images displayed on a classification page. This device can be implemented by software and / or hardware, and is generally integrated into any electronic device with network communication function, including but not limited to computers, personal digital assistants, etc. Figure 7 As shown, the image quality partial order processing device of this embodiment may include the following: information determination module 710, image quality determination module 720 and image partial order processing module 730.

[0107] The information determination module 710 is used to determine the visual information and non-visual information of the image to be processed; the visual information includes image features that can be obtained through visual elements in the image to be processed, and the non-visual information includes image features obtained by parsing the non-visual elements associated with the image to be processed.

[0108] The image quality determination module 720 is used to fuse the visual information and non-visual information to perform an image quality determination operation, thereby obtaining the predicted image quality of the image to be processed.

[0109] The image partial order processing module 730 is used to perform an image partial order operation on the image to be processed based on the predicted image quality of the image to be processed.

[0110] Based on the above embodiments, optionally, the visual information is used to measure the image content attributes, which are described by at least one of the following: image color, image texture, image lines, and image content richness; the non-visual information is used to measure additional image attributes other than the image content, which are described by at least one of the following: image format, image color space type, image size, and image creation information.

[0111] Based on the above embodiments, optionally, the image quality determination module 720 includes:

[0112] The visual information of the image to be processed is input into the first image quality judgment model, and the visual features of the image to be processed are obtained by performing a visual feature extraction operation.

[0113] The non-visual features in the non-visual information are fused with the visual features and input into the second image quality judgment model to obtain the predicted image quality of the image to be processed.

[0114] The first image quality assessment model is obtained by training the model using visual information from training sample images in the training set; the second image quality assessment model is obtained by training the model using both visual and non-visual information from the training sample images.

[0115] Based on the above embodiments, optionally, both the first image quality assessment model and the second image quality assessment model are trained using the image export click ratio as an image quality assessment metric.

[0116] Based on the above embodiments, optionally, the number of image clicks of the training sample images is greater than a preset threshold, and each training sample image is evenly distributed across different image click ratios within a preset range after being sorted according to the image export click ratio, rather than being clustered within a certain image click ratio range.

[0117] Based on the above embodiments, optionally, the non-visual features in the non-visual information are fused with the visual features and input into the second image quality determination model, including:

[0118] Discretized non-visual features are obtained by encoding and mapping non-visual features in non-visual information.

[0119] Multimodal features are obtained by fusing the visual features of the image to be processed with discretized non-visual features.

[0120] The multimodal features are input into the second image quality assessment model to perform image quality assessment operations, and output the predicted image quality of the image to be processed.

[0121] Based on the above embodiments, optionally, the first image quality determination model is obtained by iterative training based at least on a visual feature extraction network.

[0122] Based on the above embodiments, optionally, the second image quality determination model is trained iteratively based at least on a gradient boosting framework network of a weak classifier.

[0123] Optionally, based on the above embodiments, the device further includes:

[0124] Determine test sample images and reference images of the test sample images from the test set;

[0125] The image quality of the test sample image and the reference image of the test sample image are determined by the first image quality determination model and the second image quality determination model, so as to obtain the image quality of the test sample image and the predicted image quality of the reference image.

[0126] The predicted image quality skewness and the pre-labeled image quality skewness between the test sample image and the reference image are statistically analyzed to determine whether to initiate an update of the first image quality judgment model and the second image quality judgment model.

[0127] Based on the above embodiments, optionally, the difference in image export click ratio between the test sample image and the reference image is greater than a preset multiple of the test standard deviation, wherein the test standard deviation is determined based on the standard deviation of the image export click ratio of each test sample image in the test set.

[0128] Based on the above embodiments, optionally, statistical analysis is performed on the predicted image quality skewness and pre-labeled image quality skewness between the test sample image and the reference image, including:

[0129] If the predicted image quality bias between the test sample image and the reference image is consistent with the pre-labeled image quality bias, then the image quality determination of the test sample image and the reference image is taken as a positive test example.

[0130] If the predicted image quality bias between the test sample image and the reference image is inconsistent with the pre-labeled image quality bias, then the image quality judgment of the test sample image and the reference image is taken as a negative test example.

[0131] The number of positive test cases and the number of negative test cases are counted, and the ratio between the number of positive test cases and the number of negative test cases is calculated as the positive-to-negative ratio.

[0132] The image quality partial order processing apparatus provided in this embodiment can execute the image quality partial order processing method provided in any of the above embodiments of this disclosure, and has the corresponding functions and beneficial effects of executing the image quality partial order processing method. For details, please refer to the relevant operations of the image quality partial order processing method in the foregoing embodiments.

[0133] The following is for reference. Figure 8 This diagram illustrates a structural schematic of an electronic device 800 suitable for implementing embodiments of the present disclosure. The terminal devices in the embodiments of the present disclosure may include, but are not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 8 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0134] like Figure 8 As shown, the electronic device 800 may include a processing device (e.g., a central processing unit, a graphics processor, etc.) 801, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 802 or a program loaded from a storage device 808 into a random access memory (RAM) 803. The RAM 803 also stores various programs and data required for the operation of the electronic device 800. The processing device 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0135] Typically, the following devices can be connected to I / O interface 805: input devices 806 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 807 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 808 including, for example, magnetic tapes, hard disks, etc.; and communication devices 809. Communication device 809 allows electronic device 800 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 8 An electronic device 800 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively.

[0136] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0137] The electronic device provided in this embodiment and the image quality partial order processing method provided in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0138] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the image quality partial order processing method shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 809, or installed from a storage device 808, or installed from a ROM 802. When the computer program is executed by the processing device 801, it performs the functions defined in the image quality partial order processing method of the embodiments of this disclosure.

[0139] This disclosure provides a computer storage medium storing a computer program that, when executed by a processor, implements the image quality partial order processing method provided in the above embodiments.

[0140] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0141] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0142] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0143] The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: determine visual and non-visual information of an image to be processed; the visual information includes image features obtainable from visual elements in the image to be processed, and the non-visual information includes image features obtained by parsing non-visual elements associated with the image to be processed; fuse the visual and non-visual information to perform an image quality determination operation to obtain a predicted image quality of the image to be processed; and perform an image partial ordering operation on the image to be processed based on the predicted image quality of the image to be processed.

[0144] Alternatively, the aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to perform the image quality partial order processing method described in any of the above embodiments.

[0145] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including but not limited to object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0146] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0147] The units described in the embodiments of this disclosure can be implemented in software or in hardware. The name of a unit does not necessarily limit the unit itself; for example, the first acquisition unit can also be described as "a unit that acquires at least two Internet Protocol addresses".

[0148] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0149] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0150] According to one or more embodiments of this disclosure, Example 1 provides an image quality partial order processing method, the method comprising:

[0151] Determine the visual and non-visual information of the image to be processed; the visual information includes image features that can be obtained from visual elements in the image to be processed, and the non-visual information includes image features obtained by parsing non-visual elements associated with the image to be processed;

[0152] The visual information and non-visual information are fused to perform an image quality determination operation, thereby obtaining the predicted image quality of the image to be processed.

[0153] A partial image ordering operation is performed on the image to be processed based on the predicted image quality.

[0154] According to one or more embodiments of this disclosure, Example 2 describes the method according to Example 1, wherein the visual information is used to measure image content attributes, which are described by at least one of the following: image color, image texture, image lines, and image content richness; and / or, the non-visual information is used to measure additional image attributes other than image content, which are described by at least one of the following: image format, image color space type, image size, and image creation information.

[0155] According to one or more embodiments of this disclosure, Example 3, based on the method described in Example 1, fuses the visual information with non-visual information to perform an image quality determination operation to obtain a predicted image quality of the image to be processed, including:

[0156] The visual information of the image to be processed is input into the first image quality judgment model, and the visual features of the image to be processed are obtained by performing a visual feature extraction operation.

[0157] The non-visual features in the non-visual information are fused with the visual features and input into the second image quality judgment model to obtain the predicted image quality of the image to be processed.

[0158] The first image quality assessment model is obtained by training the model using visual information from training sample images in the training set; the second image quality assessment model is obtained by training the model using both visual and non-visual information from the training sample images.

[0159] According to one or more embodiments of this disclosure, Example 4, based on the method described in Example 3, both the first image quality assessment model and the second image quality assessment model are trained using the image export click ratio as an image quality assessment metric.

[0160] According to one or more embodiments of this disclosure, Example 5 describes the method described in Example 3, wherein the number of image clicks of the training sample images is greater than a preset threshold, and each training sample image is evenly distributed across different image click ratios within a preset range after being sorted by the image export click ratio.

[0161] According to one or more embodiments of this disclosure, Example 6, based on the method described in Example 3, fuses non-visual features in non-visual information with the visual features and inputs the fusion into a second image quality determination model, including:

[0162] Discretized non-visual features are obtained by encoding and mapping non-visual features in non-visual information.

[0163] Multimodal features are obtained by fusing the visual features of the image to be processed with discretized non-visual features.

[0164] The multimodal features are input into the second image quality assessment model to perform image quality assessment operations, and output the predicted image quality of the image to be processed.

[0165] According to one or more embodiments of this disclosure, Example 7, according to any one of Examples 3-6, the first image quality determination model is obtained by iterative training based at least on a visual feature extraction network.

[0166] According to one or more embodiments of this disclosure, Example 8 describes the second image quality determination model iteratively trained based at least on a gradient boosting framework network of a weak classifier, according to any one of Examples 3-6.

[0167] According to one or more embodiments of this disclosure, Example 9 describes the method according to Example 3, the method further comprising:

[0168] Determine test sample images and reference images of the test sample images from the test set;

[0169] The image quality of the test sample image and the reference image of the test sample image are determined by the first image quality determination model and the second image quality determination model, so as to obtain the image quality of the test sample image and the predicted image quality of the reference image.

[0170] The predicted image quality skewness and the pre-labeled image quality skewness between the test sample image and the reference image are statistically analyzed to determine whether to initiate an update of the first image quality judgment model and the second image quality judgment model.

[0171] According to one or more embodiments of this disclosure, Example 10 describes the method of Example 9 in which the difference in image export click ratio between the test sample image and the reference image is greater than a preset multiple of the test standard deviation, the test standard deviation being determined based on the standard deviation of the image export click ratio of each test sample image in the test set.

[0172] According to one or more embodiments of this disclosure, Example 11, based on the method described in Example 9, performs statistical analysis on the predicted image quality skewness and the pre-labeled image quality skewness between the test sample image and the reference image, including:

[0173] If the predicted image quality bias between the test sample image and the reference image is consistent with the pre-labeled image quality bias, then the image quality determination of the test sample image and the reference image is taken as a positive test example.

[0174] If the predicted image quality bias between the test sample image and the reference image is inconsistent with the pre-labeled image quality bias, then the image quality judgment of the test sample image and the reference image is taken as a negative test example.

[0175] The number of positive test cases and the number of negative test cases are counted, and the ratio between the number of positive test cases and the number of negative test cases is calculated as the positive-to-negative ratio.

[0176] According to one or more embodiments of this disclosure, Example 12 provides an image quality partial order processing apparatus, the apparatus comprising:

[0177] The information determination module is used to determine the visual information and non-visual information of the image to be processed; the visual information includes image features that can be obtained through visual elements in the image to be processed, and the non-visual information includes image features obtained by parsing non-visual elements associated with the image to be processed;

[0178] The image quality determination module is used to fuse the visual information and non-visual information to perform an image quality determination operation and obtain the predicted image quality of the image to be processed.

[0179] The image partial order processing module is used to perform image partial ordering operations on the image to be processed based on the predicted image quality of the image to be processed.

[0180] According to one or more embodiments of this disclosure, Example 13 provides an electronic device, the electronic device comprising:

[0181] At least one processor; and

[0182] A memory communicatively connected to the at least one processor; wherein,

[0183] The memory stores a computer program that can be executed by the at least one processor to enable the at least one processor to perform the image quality partial order processing method described in any one of Examples 1-11.

[0184] According to one or more embodiments of the present disclosure, Example 14 provides a computer-readable medium storing computer instructions for causing a processor to execute the image quality partial order processing method described in any one of Examples 1-11.

[0185] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

[0186] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0187] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A method for processing image quality partial order, characterized in that, The method includes: Determine the visual and non-visual information of the image to be processed; the visual information includes image features that can be obtained from visual elements in the image to be processed, and the non-visual information includes image features obtained by parsing non-visual elements associated with the image to be processed; The visual information and non-visual information are fused to perform an image quality determination operation, thereby obtaining the predicted image quality of the image to be processed. A partial image ordering operation is performed on the image to be processed based on the predicted image quality. The process of fusing visual and non-visual information to perform an image quality determination operation, thereby obtaining the predicted image quality of the image to be processed, includes: The visual information of the image to be processed is input into the first image quality judgment model, and the visual features of the image to be processed are obtained by performing a visual feature extraction operation. The non-visual features obtained from the non-visual information are fused with the visual features and input into the second image quality judgment model to obtain the predicted image quality of the image to be processed. The first image quality assessment model is obtained by training the model using visual information from the training sample images in the training set; the second image quality assessment model is obtained by training the model using both visual and non-visual information from the training sample images; the number of image clicks on the training sample images is greater than a preset threshold, and each training sample image is evenly distributed across different image click ratios within a preset range after being sorted by the image export click ratio.

2. The method according to claim 1, characterized in that, The visual information is used to measure the image content attributes, which are described by at least one of the following: image color, image texture, image lines, and image content richness; and / or The non-visual information is used to measure additional image attributes other than the image content. These additional image attributes are described by at least one of the following: image format, image color space type, image size, and image creation information.

3. The method according to claim 1, characterized in that, Both the first image quality assessment model and the second image quality assessment model use the image export click ratio as an image quality assessment metric during model training.

4. The method according to claim 1, characterized in that, The non-visual features in the non-visual information are fused with the visual features and input into the second image quality assessment model, including: Discretized non-visual features are obtained by encoding and mapping non-visual features in non-visual information. Multimodal features are obtained by fusing the visual features of the image to be processed with discretized non-visual features. The multimodal features are input into the second image quality assessment model to perform image quality assessment operations, and output the predicted image quality of the image to be processed.

5. The method according to any one of claims 1, 3, and 4, characterized in that, The first image quality assessment model is obtained through iterative training based on a visual feature extraction network.

6. The method according to any one of claims 1, 3, and 4, characterized in that, The second image quality assessment model is trained iteratively based on a gradient boosting framework network with a weak classifier.

7. The method according to claim 1, characterized in that, The method further includes: Determine test sample images and reference images of the test sample images from the test set; The image quality of the test sample image and the reference image of the test sample image are determined by the first image quality determination model and the second image quality determination model, so as to obtain the image quality of the test sample image and the predicted image quality of the reference image. The predicted image quality skewness and the pre-labeled image quality skewness between the test sample image and the reference image are statistically analyzed to determine whether to initiate an update of the first image quality judgment model and the second image quality judgment model.

8. The method according to claim 7, characterized in that, The difference in the image export click ratio between the test sample image and the reference image is greater than a preset multiple of the test standard deviation, which is determined based on the standard deviation of the image export click ratio of each test sample image in the test set.

9. The method according to claim 7, characterized in that, Statistical analysis is performed on the predicted image quality skewness and pre-labeled image quality skewness between the test sample image and the reference image, including: If the predicted image quality bias between the test sample image and the reference image is consistent with the pre-labeled image quality bias, then the image quality determination of the test sample image and the reference image is taken as a positive test example. If the predicted image quality bias between the test sample image and the reference image is inconsistent with the pre-labeled image quality bias, then the image quality judgment of the test sample image and the reference image is taken as a negative test example. The number of positive test cases and the number of negative test cases are counted, and the ratio between the number of positive test cases and the number of negative test cases is calculated as the positive-to-negative ratio.

10. An image quality partial order processing device, characterized in that, The device includes: The information determination module is used to determine the visual information and non-visual information of the image to be processed; the visual information includes image features that can be obtained through visual elements in the image to be processed, and the non-visual information includes image features obtained by parsing non-visual elements associated with the image to be processed; The image quality determination module is used to fuse the visual information and non-visual information to perform an image quality determination operation and obtain the predicted image quality of the image to be processed. The image partial order processing module is used to perform image partial ordering operations on the image to be processed based on the predicted image quality of the image to be processed. The image quality assessment module includes: The visual information of the image to be processed is input into the first image quality judgment model, and the visual features of the image to be processed are obtained by performing a visual feature extraction operation. The non-visual features in the non-visual information are fused with the visual features and input into the second image quality judgment model to obtain the predicted image quality of the image to be processed. The first image quality assessment model is obtained by training the model using visual information from training sample images in the training set; the second image quality assessment model is obtained by training the model using both visual and non-visual information from the training sample images; the number of image clicks of the training sample images is greater than a preset threshold, and each training sample image is evenly distributed across different image click ratios within a preset range after being sorted by the image export click ratio, rather than clustered within a certain image click ratio range.

11. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the image quality partial order processing method according to any one of claims 1-9.

12. A computer-readable medium, characterized in that, The computer-readable medium stores computer instructions that cause a processor to execute the image quality partial order processing method according to any one of claims 1-9.

Images