Information processing method and device, program product and electronic equipment

Abstract

The invention provides an information processing method and device, a program product and electronic equipment, and relates to the technical field of computers. The method comprises the steps of determining a to-be-evaluated image; performing multi-dimensional feature extraction processing on the to-be-evaluated image to obtain fusion features; the fusion feature is correspondingly determined based on at least two of an object feature, a saturation feature, a composition feature, a content plumpness feature and a definition feature; and according to the fusion feature and a quality evaluation rule, determining a quality evaluation value of the to-be-evaluated image, so as to perform publishing processing on the to-be-evaluated image according to the quality evaluation value. The image is comprehensively evaluated from multiple dimensions of portrait, saturation, composition, content fullness and the like, and the accuracy of visual quality evaluation of the image is improved.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the field of computer technology, and more particularly, to an information processing method, an information processing device, a computer program product, and an electronic device. Background Art

[0002] At present, in today's digital age, images have become an important carrier of information dissemination and expression, and are widely used in many fields such as social media, advertising design, photography art, film and television production, etc.

[0003] In recent years, in order to share attractive or interesting images with other users, users or companies can often post the images they take to social media platforms to increase the dissemination and influence of the images. Summary of the Invention

[0004] However, with the explosive growth in the number of images, how to quickly and accurately evaluate the visual quality of images and improve their dissemination and influence has become an urgent problem to be solved.

[0005] In view of this, the present disclosure provides an information processing method, an information processing device, a computer program product, and an electronic device to improve the accuracy of evaluating the visual quality of an image to a certain extent, thereby enhancing the dissemination and influence of the image.

[0006] According to a first aspect of the present disclosure, there is provided an information processing method, the method comprising:

[0007] Determine the image to be evaluated;

[0008] Performing multi-dimensional feature extraction processing on the image to be evaluated to obtain a fusion feature; the fusion feature is determined based on at least two of the object feature, saturation feature, composition feature, content fullness feature, and clarity feature;

[0009] The quality evaluation value of the image to be evaluated is determined according to the fusion features and the quality evaluation rule, so as to perform publishing processing on the image to be evaluated according to the quality evaluation value.

[0010] In a possible implementation, performing multi-dimensional feature extraction processing on the image to be evaluated to obtain fusion features includes:

[0011] Performing feature extraction processing on the image to be evaluated using the first model to obtain corresponding features; wherein the feature extraction processing includes any one of object feature extraction processing, saturation feature extraction processing, composition feature extraction processing, content fullness feature extraction processing, and clarity feature extraction processing;

[0012] The corresponding features obtained are fused using the first model to obtain the fused features.

[0013] In a possible implementation, if the feature extraction process is an object feature extraction process, performing feature extraction on the image to be evaluated using the first model to obtain corresponding features includes:

[0014] determining whether the image to be evaluated includes a portrait;

[0015] When it is determined that the image to be evaluated includes a portrait, performing portrait feature extraction processing on the image to be evaluated using the multi-task branch model in the first model to obtain multiple portrait sub-features; wherein the multiple portrait sub-features include at least two of the following: body sub-features, face sub-features, gender sub-features, clothing category sub-features, and face quality sub-features;

[0016] A portrait feature is determined based on the multiple portrait sub-features.

[0017] In a possible implementation, if the feature extraction process is a saturation feature extraction process, performing feature extraction on the image to be evaluated using the first model to obtain corresponding features includes:

[0018] Performing color conversion on the image to be evaluated to obtain a converted first image;

[0019] The convolutional neural network model in the first model is used to perform feature extraction processing on the first image to obtain a saturation feature corresponding to the image to be evaluated.

[0020] In a possible implementation, if the feature extraction process is a composition feature extraction process, performing feature extraction on the image to be evaluated using the first model to obtain corresponding features includes:

[0021] Performing global feature extraction processing on the image to be evaluated to obtain a first composition feature;

[0022] Performing local feature extraction processing on the image to be evaluated to obtain a second composition feature;

[0023] The composition feature is determined according to the first composition feature and the second composition feature.

[0024] In a possible implementation, performing global feature extraction processing on the image to be evaluated to obtain a first composition feature includes:

[0025] Performing global feature extraction processing on the image to be evaluated through the transformer model in the first model to obtain a first composition feature.

[0026] In a possible implementation, performing local feature extraction processing on the image to be evaluated to obtain the second composition feature includes:

[0027] Performing edge detection processing on the image to be evaluated to obtain edge detection information of the image to be evaluated;

[0028] The edge detection information is input into the convolutional neural network model in the first model to obtain a second composition feature.

[0029] In a possible implementation, if the feature extraction process is content fullness feature extraction process, then performing feature extraction on the image to be evaluated using the first model to obtain corresponding features includes:

[0030] Determining the content and location information contained in the image to be evaluated through the multi-target detection and multi-label model in the first model;

[0031] Determine the content fullness feature of the image to be evaluated based on the content and position information.

[0032] In a possible implementation, if the feature extraction process is a clarity feature extraction process, performing feature extraction on the image to be evaluated using the first model to obtain corresponding features includes:

[0033] Determining the clarity information corresponding to the image to be evaluated by using the physical feature extraction model in the first model;

[0034] Determine the clarity feature of the image to be evaluated based on the clarity information.

[0035] In a possible implementation, fusing the corresponding features obtained by the first model to obtain the fused features includes:

[0036] Determining the weight of each corresponding feature according to the contribution of each corresponding feature to the quality of the image to be evaluated;

[0037] The weights of the corresponding features and the corresponding features are processed through the fusion model based on the attention mechanism in the first model to determine the fusion feature.

[0038] In a possible implementation, the first model is trained in the following manner:

[0039] Determining a plurality of images containing a plurality of features, and labeling each feature and a quality assessment value in the plurality of images containing the plurality of features to obtain training sample images;

[0040] Inputting the training sample image into a preset first model for training to obtain a plurality of prediction values, and comparing the prediction values ​​with the corresponding labeled values ​​in the training sample image to obtain a plurality of comparison results;

[0041] The preset first model is adjusted according to the multiple comparison results to obtain the first model.

[0042] In a possible implementation, adjusting the preset first model according to the multiple comparison results to obtain the first model includes:

[0043] Determining an overall loss function based on the multiple comparison results;

[0044] According to the overall loss function, the preset first model is adjusted to obtain the first model.

[0045] In one possible implementation, determining the overall loss function includes:

[0046] Determining first prediction values ​​of a plurality of portrait sub-features detected by the preset first model, and obtaining a first loss function according to a difference between the first prediction value and a first labeled value corresponding to the first prediction value;

[0047] Determining a second predicted value of a plurality of content fullness features detected by the preset first model, and obtaining a second loss function according to a difference between the second predicted value of the content fullness and a second labeled value corresponding to the second predicted value; and

[0048] Determining a quality assessment prediction value determined by the preset first model, and obtaining a third loss function according to a difference between the quality assessment prediction value and a quality assessment annotation value corresponding to the quality assessment prediction value;

[0049] The first loss function, the second loss function and the third loss function are weightedly calculated to obtain an overall loss function.

[0050] According to a second aspect of the present disclosure, there is provided an information processing device, the device comprising:

[0051] a determination unit, configured to determine an image to be evaluated;

[0052] an extraction unit, configured to perform multi-dimensional feature extraction processing on the image to be evaluated to obtain a fusion feature; the fusion feature is determined based on at least two of the object feature, saturation feature, composition feature, content fullness feature, and clarity feature;

[0053] A processing unit is used to determine a quality evaluation value of the image to be evaluated according to the fusion feature and the quality evaluation rule, so as to perform publishing processing on the image to be evaluated according to the quality evaluation value.

[0054] In a possible implementation, the extraction unit is specifically configured to:

[0055] Performing feature extraction processing on the image to be evaluated using the first model to obtain corresponding features; wherein the feature extraction processing includes any one of object feature extraction processing, saturation feature extraction processing, composition feature extraction processing, content fullness feature extraction processing, and clarity feature extraction processing;

[0056] The corresponding features obtained are fused using the first model to obtain the fused features.

[0057] In a possible implementation, if the feature extraction process is an object feature extraction process, the extraction unit is specifically configured to:

[0058] determining whether the image to be evaluated includes a portrait;

[0059] When it is determined that the image to be evaluated includes a portrait, performing portrait feature extraction processing on the image to be evaluated using the multi-task branch model in the first model to obtain multiple portrait sub-features; wherein the multiple portrait sub-features include at least two of the following: body sub-features, face sub-features, gender sub-features, clothing category sub-features, and face quality sub-features;

[0060] A portrait feature is determined based on the multiple portrait sub-features.

[0061] In a possible implementation, if the feature extraction process is a saturation feature extraction process, the extraction unit is specifically configured to:

[0062] Performing color conversion on the image to be evaluated to obtain a converted first image;

[0063] The convolutional neural network model in the first model is used to perform feature extraction processing on the first image to obtain a saturation feature corresponding to the image to be evaluated.

[0064] In a possible implementation, if the feature extraction process is a composition feature extraction process, the extraction unit is specifically configured to:

[0065] Performing global feature extraction processing on the image to be evaluated to obtain a first composition feature;

[0066] Performing local feature extraction processing on the image to be evaluated to obtain a second composition feature;

[0067] The composition feature is determined according to the first composition feature and the second composition feature.

[0068] In a possible implementation, the extraction unit is specifically configured to:

[0069] Performing global feature extraction processing on the image to be evaluated through the transformer model in the first model to obtain a first composition feature.

[0070] In a possible implementation, the extraction unit is specifically configured to:

[0071] Performing edge detection processing on the image to be evaluated to obtain edge detection information of the image to be evaluated;

[0072] The edge detection information is input into the convolutional neural network model in the first model to obtain a second composition feature.

[0073] In a possible implementation, if the feature extraction process is content fullness feature extraction process, the extraction unit is specifically configured to:

[0074] Determining the content and location information contained in the image to be evaluated through the multi-target detection and multi-label model in the first model;

[0075] Determine the content fullness feature of the image to be evaluated based on the content and position information.

[0076] In a possible implementation, if the feature extraction process is a clarity feature extraction process, the extraction unit is specifically configured to:

[0077] Determining the clarity information corresponding to the image to be evaluated by using the physical feature extraction model in the first model;

[0078] Determine the clarity feature of the image to be evaluated based on the clarity information.

[0079] In a possible implementation, the extraction unit is specifically configured to:

[0080] Determining the weight of each corresponding feature according to the contribution of each corresponding feature to the quality of the image to be evaluated;

[0081] The weights of the corresponding features and the corresponding features are processed through the fusion model based on the attention mechanism in the first model to determine the fusion feature.

[0082] In one possible implementation, the apparatus includes a training unit configured to:

[0083] Determining a plurality of images containing a plurality of features, and labeling each feature and a quality assessment value in the plurality of images containing the plurality of features to obtain training sample images;

[0084] Inputting the training sample image into a preset first model for training to obtain a plurality of prediction values, and comparing the prediction values ​​with the corresponding labeled values ​​in the training sample image to obtain a plurality of comparison results;

[0085] The preset first model is adjusted according to the multiple comparison results to obtain the first model.

[0086] In a possible implementation, the training unit is configured to:

[0087] Determining an overall loss function based on the multiple comparison results;

[0088] According to the overall loss function, the preset first model is adjusted to obtain the first model.

[0089] In a possible implementation, the training unit is configured to:

[0090] Determining first prediction values ​​of a plurality of portrait sub-features detected by the preset first model, and obtaining a first loss function according to a difference between the first prediction value and a first labeled value corresponding to the first prediction value;

[0091] Determining a second predicted value of a plurality of content fullness features detected by the preset first model, and obtaining a second loss function according to a difference between the second predicted value of the content fullness and a second labeled value corresponding to the second predicted value; and

[0092] Determining a quality assessment prediction value determined by the preset first model, and obtaining a third loss function according to a difference between the quality assessment prediction value and a quality assessment annotation value corresponding to the quality assessment prediction value;

[0093] The first loss function, the second loss function and the third loss function are weightedly calculated to obtain an overall loss function.

[0094] According to a third aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the method of the first aspect and possible implementations thereof are implemented.

[0095] According to a fourth aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the method of the above-mentioned first aspect and its possible implementation methods by executing the executable instructions.

[0096] The technical solution disclosed in this disclosure has the following beneficial effects:

[0097] In the embodiment of the present disclosure, the image to be evaluated can be determined first, and then multi-dimensional feature extraction can be performed on the image to be evaluated, such as object feature extraction, saturation feature extraction, composition feature extraction, and content fullness feature extraction, and then the extracted features can be fused to obtain fusion features, that is, fusion features determined based on at least two of the object features, saturation features, composition features, content fullness features, and clarity features. Furthermore, the quality assessment value of the image to be evaluated can be determined based on the fusion features and quality assessment rules, so that the image to be evaluated can be published and processed based on the quality assessment value. It can be seen that in the embodiment of the present disclosure, the image to be evaluated can be comprehensively and carefully evaluated from multiple dimensions such as portrait, saturation, composition, and content fullness, as well as the interaction between multiple dimensions, to improve the accuracy of the visual instructions of the evaluated image.

[0098] Other features and advantages of the present disclosure will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present disclosure. The purposes and other advantages of the present disclosure can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0099] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following briefly introduces the drawings required for use in the embodiments of the present disclosure. Obviously, the drawings introduced below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0100] Figure 1 A schematic diagram of an application scenario in this exemplary embodiment is shown;

[0101] Figure 2 A schematic flow chart showing an information processing method in this exemplary embodiment;

[0102] Figure 3 A schematic diagram of a process for obtaining fusion features in this exemplary embodiment is shown;

[0103] Figure 4 A schematic diagram of a process for training a first model in this exemplary embodiment is shown;

[0104] Figure 5 A schematic diagram illustrating a method of determining a portrait feature in this exemplary embodiment is shown;

[0105] Figure 6 A schematic diagram showing a method for determining a saturation characteristic in this exemplary embodiment is shown;

[0106] Figure 7 A schematic diagram showing a method for determining a patterning feature in this exemplary embodiment is shown;

[0107] Figure 8 A schematic diagram illustrating a method for determining content fullness characteristics in this exemplary embodiment;

[0108] Figure 9 A schematic structural diagram of an information processing device according to this exemplary embodiment is shown;

[0109] Figure 10 A schematic structural diagram of an electronic device in this exemplary embodiment is shown. DETAILED DESCRIPTION

[0110] In order to make the purpose, technical solutions and advantages of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present disclosure. In the absence of conflict, the embodiments in the present disclosure and the features in the embodiments can be arbitrarily combined with each other. In addition, although the logical order is shown in the flow chart, in some cases, the steps shown or described can be performed in an order different from that here.

[0111] The terms "comprises" and "comprising" and any variations thereof in the specification and claims of this disclosure are intended to cover non-exclusive protection. For example, a process, method, system, product, or apparatus comprising a series of steps or elements is not limited to the listed steps or elements, but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or apparatus.

[0112] In the embodiments of the present disclosure, one or more, "many" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: the existence of A alone, the existence of A and B at the same time, and the existence of B alone, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or plural.

[0113] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order, sequence, size and priority. For example, the first semantic vector and the second semantic vector in the embodiment of the present disclosure are only used to distinguish different semantic vectors. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of apparatus and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0114] The following describes exemplary embodiments of the present disclosure in conjunction with the accompanying drawings. The accompanying drawings are schematic diagrams of the present disclosure and are not necessarily drawn to scale. Some of the block diagrams shown in the accompanying drawings may be functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, or in hardware modules or integrated circuits, or in networks, processors or microcontrollers. The embodiments can be implemented in various forms and should not be construed as being limited to the examples set forth herein. The features, structures or characteristics described in the present disclosure can be combined in one or more embodiments in any suitable manner. In the description below, many specific details are provided to provide a full description of the embodiments of the present disclosure. However, those skilled in the art will appreciate that one or more specific details may be omitted when implementing the technical solutions of the present disclosure, or that other methods, components, devices, steps, etc. may be used to replace one or more specific details.

[0115] It should be noted that in the embodiments of this disclosure, certain software, components, models, and other existing industry solutions may be mentioned. These should be considered exemplary and are intended solely to illustrate the feasibility of implementing the technical solutions of this disclosure. This does not mean that the applicant has or will necessarily use such solutions. The collection, dissemination, use, and display of data in this technical solution comply with relevant national laws and regulations.

[0116] Application Scenario Overview

[0117] To better understand the technical solutions provided by the embodiments of the present disclosure, the following briefly introduces the application scenarios to which the technical solutions provided by the embodiments of the present disclosure are applicable. It should be noted that the application scenarios introduced below are only used to illustrate the embodiments of the present disclosure and are not limiting. In specific implementations, the technical solutions provided by the embodiments of the present disclosure can be flexibly applied according to actual needs.

[0118] In the embodiments of the present disclosure, information processing technology can be applied to various business scenarios that provide image quality evaluation services to users, such as business scenarios for image quality evaluation of photos posted by users on social media, or business scenarios for image quality evaluation in advertising design business, or business scenarios for image quality evaluation in film and television production. This is not limited in the embodiments of the present disclosure.

[0119] See also Figure 1 As shown, Figure 1 This is an application scenario to which the technical solution of the embodiment of the present disclosure can be applied. In the scenario diagram, it includes a terminal device 110 and an electronic device 120. There can be one or more terminal devices 110, and each terminal device 110 can be provided with an information processing system platform, so that different users can log in to the information processing system platform deployed on the corresponding terminal device 110. Figure 1 In the figure, a terminal device 110 is used as an example. The electronic device 120 can be understood as a device that provides services to the information processing system. In addition, the terminal device 110 and the electronic device 120 communicate with each other via one or more networks 130.

[0120] In an embodiment of the present disclosure, a user can log in to the platform of the information processing system deployed on the corresponding terminal device 110 and determine the image to be evaluated. In this way, the terminal device 110 can send the image to be evaluated to the electronic device 120, so that the electronic device 120 can receive the image to be evaluated, and then perform multi-dimensional feature extraction processing on the image to be evaluated to obtain a fusion feature; the fusion feature is determined based on at least two of the object feature, saturation feature, composition feature, content fullness feature and clarity feature; according to the fusion feature and the quality evaluation rule, the quality evaluation value of the image to be evaluated is determined, so that the image to be evaluated is published according to the quality evaluation value.

[0121] In the embodiments of the present disclosure, Figure 1 The terminal device 110 can be a mobile phone, a tablet computer (PAD), a personal computer, a smart TV, a smart watch, a smart speaker, a smart car device, and a wearable device, but is not limited thereto.

[0122] In the embodiments of the present disclosure, Figure 1 The electronic device 120 may also be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server or cloud server cluster that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms, but is not limited to these.

[0123] Of course, the method provided in the embodiment of the present disclosure is not limited to Figure 1 The application scenario shown can also be used in other possible application scenarios, for example, only the electronic device 120 executes the information processing method, which is not limited in the embodiment of the present disclosure.

[0124] Exemplary Methods

[0125] To further illustrate the technical solutions provided by the embodiments of the present disclosure, this is described in detail below in conjunction with the accompanying drawings and specific implementation methods. Although the embodiments of the present disclosure provide the method operation steps as shown in the following embodiments or drawings, more or fewer operation steps may be included in the method based on routine or no creative labor. In steps where there is no necessary causal relationship logically, the execution order of these steps is not limited to the execution order provided by the embodiments of the present disclosure. In the actual processing process or when the device is executed, the method may be executed in the order of the methods shown in the embodiments or drawings or in parallel.

[0126] See Figure 2 , Figure 2 FIG is a flow chart of an information processing method according to an embodiment of the present disclosure. The flow of the method may be executed by an electronic device, which may be Figure 1 The electronic device 120 in the embodiment of the present invention is implemented as follows:

[0127] Step 201: Determine an image to be evaluated.

[0128] In the embodiment of the present disclosure, the electronic device may select the image to be evaluated from the database based on the image screening rule, or the electronic device may receive the image to be evaluated sent by the terminal device, which is not limited in the embodiment of the present disclosure.

[0129] Step 202: Perform multi-dimensional feature extraction processing on the image to be evaluated to obtain fusion features; the fusion features are determined based on at least two of the object features, saturation features, composition features, content fullness features, and clarity features.

[0130] In the embodiments of the present disclosure, see Figure 3 , the electronic device may determine the fusion feature by adopting but not limited to the following steps:

[0131] Step 301: Perform feature extraction processing on the image to be evaluated through the first model to obtain corresponding features; wherein the feature extraction processing includes any one of object feature extraction processing, saturation feature extraction processing, composition feature extraction processing, content fullness feature extraction processing and clarity feature extraction processing.

[0132] In the embodiment of the present disclosure, before introducing a solution for performing feature extraction processing on the image to be evaluated by using the first model, a process of obtaining the first model is first introduced.

[0133] See also Figure 4 As shown, Figure 4 Schematic diagram of the process of training the first model in the embodiment of the present disclosure.

[0134] Step 401: Determine a plurality of images containing a plurality of features, and label each feature and quality assessment value in the plurality of images containing a plurality of features to obtain training sample images.

[0135] In the disclosed embodiment, the multiple features may be a portrait dimension feature, a saturation dimension feature, a composition dimension feature, a content richness dimension feature, and a clarity dimension feature. In other words, each training sample image is annotated with a quality assessment value for the image and the feature information values ​​of the multiple features in the image.

[0136] Step 402: Input the training sample image into a preset first model for training to obtain a plurality of prediction values, and compare the prediction values ​​with the labeled values ​​in the corresponding training sample image to obtain a plurality of comparison results.

[0137] In the disclosed embodiment, each training sample image is associated with a prediction value, which includes a quality assessment prediction value and feature prediction values ​​corresponding to a plurality of features. This allows the quality assessment prediction value to be compared with the quality assessment label value of the image, and the label values ​​corresponding to the plurality of features to be compared with the feature prediction values, thereby obtaining a comparison result for the training sample image. Thus, when a preset first model is trained using multiple training sample images, multiple comparison results corresponding to the multiple training sample images can be obtained.

[0138] Step 403: Adjust the preset first model according to the multiple comparison results to obtain the first model.

[0139] In an embodiment of the present disclosure, an overall loss function may be determined based on multiple comparison results; and then, the preset first model may be adjusted based on the overall loss function to obtain the first model.

[0140] In the embodiment of the present disclosure, the overall loss function may be determined by, but not limited to, the following steps:

[0141] Step A: Determine a first prediction value of a plurality of portrait sub-features detected by a preset first model, and obtain a first loss function according to a difference between the first prediction value and a first labeled value corresponding to the first prediction value.

[0142] In the disclosed embodiment, the first loss function corresponding to the human body-related features can be determined based on the sum of the loss information corresponding to whether the training sample image contains a face, the loss information corresponding to the face label, the loss information corresponding to the face gender, the loss information corresponding to the face quality, the loss information corresponding to the clothing type, the loss information corresponding to the body frame, and the loss information corresponding to the face skeleton point. In other words, the portrait sub-features can include whether the face is contained, the facial expression, the face gender, the face quality, the clothing type, the body frame, the face skeleton point, etc.

[0143] Step B: Determine a second prediction value of detecting multiple content fullness features through a preset first model, and obtain a second loss function based on the difference between the second prediction value of the content fullness and a second labeled value corresponding to the second prediction value.

[0144] In an embodiment of the present disclosure, the second loss function can be determined according to the loss information of the target detection box in the training sample image and the loss information of the target classification.

[0145] Step C: Determine the quality assessment prediction value determined by the preset first model, and obtain a third loss function according to the difference between the quality assessment prediction value and the quality assessment annotation value corresponding to the quality assessment prediction value.

[0146] In the embodiment of the present disclosure, the third loss function may be determined based on the finger quality evaluation value.

[0147] Step D: Perform weighted calculation on the first loss function, the second loss function, and the third loss function to obtain the overall loss function.

[0148] In the disclosed embodiments, for ease of description, the first loss function may be denoted as loss_face, the second loss function as loss_object, and the third loss function as loss_quality. Thus, the overall loss function may be determined based on the first, second, and third loss functions. The overall loss function may be denoted as Loss = a*loss_quality + b*loss_face + c*loss_object.

[0149] In the embodiment of the present disclosure, in order to ensure the effectiveness of training and features of each dimension, targeted adjustments can be made to features of different dimensions at different stages during the training process, that is, the ratio of weight coefficients a, b, and c is adjusted, so as to ensure the effectiveness and pertinence of training at different stages as much as possible.

[0150] In the embodiment of the present disclosure, in order to fully retain the role of dimensional features during the training process, a joint loss function training scheme is designed, so that the preset first model is trained based on the overall loss function to obtain the trained first model.

[0151] In the embodiment of the present disclosure, after the first model is obtained, feature extraction processing can be performed on the image to be evaluated using the first model to obtain corresponding features.

[0152] In one possible embodiment, see, for example, Figure 5 As shown, if the feature extraction processing is object feature extraction processing, the electronic device determines whether the image to be evaluated includes a portrait; when it is determined that the image to be evaluated includes a portrait, the multi-task branch model in the first model is used to perform portrait feature extraction processing on the image to be evaluated to obtain multiple portrait sub-features; wherein the multiple portrait sub-features include at least two of human body sub-features, face sub-features, gender sub-features, clothing category sub-features and face image quality sub-features; based on the multiple portrait sub-features, the portrait features are determined.

[0153] Optionally, the multi-task branch model can be designed with a deep convolutional neural network (Residual Network 101, resnet101) as the backbone network.

[0154] It can be seen that in the embodiment of the present disclosure, when there is a portrait in the image to be evaluated, the posture of the portrait, the proportion of the portrait in the image, and the attributes of the portrait are taken into consideration; among which, the attributes of the portrait include facial gender, facial expression, facial appearance, body clothing type, etc., all of which have a great influence on the overall beauty of the picture. Therefore, in terms of portrait dimension features, the present disclosure adopts a multi-task branch model to uniformly identify whether there is a portrait, a portrait frame, whether there is a face, a face frame, facial attributes, body posture, clothing type, etc. in the image to be evaluated, so as to extract the aforementioned features in the image, that is, obtain the portrait feature, which can be recorded as F1.

[0155] In one possible embodiment, see, for example, Figure 6 As shown, if the feature extraction processing is saturation feature extraction processing, the image to be evaluated is subjected to color conversion processing to obtain the converted first image; the convolutional neural network model in the first model is used to perform feature extraction processing on the first image to obtain the saturation feature corresponding to the image to be evaluated.

[0156] In the embodiment of the present disclosure, it is considered that after converting the image to be evaluated into the HSV color space, the saturation test can be performed more quickly and accurately. Therefore, when calculating the saturation dimension feature of the image to be evaluated, the present disclosure can first convert the RGB image (i.e., the aforementioned image to be evaluated) into the HSV color space. Considering that the color texture does not require too deep features, when extracting the saturation dimension feature, the image in the HSV space can only be input into a shallow conventional convolutional neural network (CNN) (i.e., the aforementioned convolutional neural network model), so that the feature vector can be extracted, that is, the saturation feature corresponding to the image to be evaluated. The saturation feature corresponding to the image to be evaluated can be recorded as F2.

[0157] In one possible embodiment, see, for example, Figure 7 As shown, if the feature extraction process is a composition feature extraction process, a global feature extraction process is performed on the image to be evaluated to obtain a first composition feature; a local feature extraction process is performed on the image to be evaluated to obtain a second composition feature; and the composition feature is determined based on the first composition feature and the second composition feature.

[0158] In the embodiments disclosed herein, it is considered that the composition of an image determines the layout and combination of the various elements in the image, which has an important impact on the visual effect and aesthetic value of the image. In the composition of an image, the key composition elements involved are: symmetry, which is used to analyze the structural similarity of the left and right / upper and lower halves of the image. The rule of thirds and the golden section are used to analyze and detect whether the subject in the image is located at the intersection of the nine-square grid or the golden ratio point. Line guidance, which extracts the direction and distribution of the dominant lines by analyzing the edge lines of the image. Foreground-background hierarchy, which locates the main area to analyze its proportional relationship with the background. Balance and white space are used to analyze and calculate the density difference of the element distribution and the proportion of blank areas. The above-mentioned key elements are used to evaluate the image features of the composition dimension. Therefore, based on the aforementioned composition elements, in the embodiments disclosed herein, when extracting the composition dimension features, a dual-stream network structure is designed to extract the global features and local features of the image respectively, and effectively fuse the global and local features.

[0159] Optionally, a global feature extraction process may be performed on the image to be evaluated by the transformer model in the first model to obtain the first composition feature.

[0160] In the embodiment of the present disclosure, the Transformer model is based on the attention mechanism and can pay weighted attention to different areas in the image, so as to better capture the global structural information of the image. In the composition analysis of the image to be evaluated, the image to be evaluated can be divided into multiple small blocks, and each small block is input into the Transformer model as an input sequence. The Transformer model then calculates the degree of correlation between each small block and other small blocks through the attention mechanism, thereby learning the composition features such as the distribution of elements in the image, the center of gravity position, and the spatial hierarchy. That is to say, in a scene image containing multiple objects, the Transformer model can pay attention to the spatial relationship between different objects and their layout in the picture, thereby extracting the global composition feature vector of the image, that is, the aforementioned first composition feature, which can be denoted as F3_1.

[0161] Optionally, edge detection processing may be performed on the image to be evaluated to obtain edge detection information of the image to be evaluated; the edge detection information may be input into a convolutional neural network model in the first model to obtain a second composition feature.

[0162] In the embodiment of the present disclosure, the image to be evaluated can be extracted using an edge detection algorithm, and the extracted edge map can be used as the input of a simple CNN shallow network to extract the second composition feature of the image, that is, the local composition feature, and the second composition feature is recorded as F3_2.

[0163] Optionally, the edge detection algorithm is, for example, an algorithm based on first-order derivatives (such as Roberts operator edge detection, Prewitt operator edge detection, Sobel operator edge detection, etc.), or an algorithm based on second-order derivatives (such as Laplacian operator, etc.), or a comprehensive algorithm (such as Canny edge detection algorithm), etc., which is not limited in the embodiments of the present disclosure.

[0164] In the embodiment of the present disclosure, after the above-mentioned F3_1 and F3_2 are fused in the feature channel dimension, the feature of the composition dimension of the entire image, namely the composition feature, is recorded as F3.

[0165] In one possible embodiment, see, for example, Figure 8 As shown, if the feature extraction processing is content fullness feature extraction processing, the content and position information contained in the image to be evaluated are determined through the multi-target detection and multi-label model in the first model; based on the content and position information, the content fullness characteristics of the image to be evaluated are determined.

[0166] In the disclosed embodiments, the content richness of an image can be evaluated from two key aspects: the richness of image elements and the rationality of their distribution. The richness of image elements can be understood as the types of objects contained in the image, while the rationality of their distribution can be understood as the relationships between the types of objects.

[0167] In the disclosed embodiment, multi-target detection and a multi-label model can be used to extract information such as the content and distribution of elements in the image to be evaluated, thereby determining the content richness feature of the image to be evaluated. Optionally, the YOLO11 (You Only Look Once 11) framework can be used as a feature extraction module to detect and locate the types of objects in the image, while predicting the target content and location information contained in the image, thereby extracting the content richness feature, which is denoted as F4.

[0168] In a possible implementation, if the feature extraction process is clarity feature extraction process, clarity information corresponding to the image to be evaluated is determined through the physical feature extraction model in the first model; and clarity features of the image to be evaluated are determined based on the clarity information.

[0169] In the disclosed embodiment, a physical feature extraction model can be used to determine the clarity information corresponding to the image to be evaluated. Based on the clarity information, a clarity feature of the image to be evaluated is determined, and the clarity feature is recorded as F5. Optionally, the physical feature extraction model can be understood as a network design using a transformer model or a CNN network as the backbone network.

[0170] Step 302: The corresponding features obtained are fused using the first model to obtain fused features.

[0171] In the embodiment of the present disclosure, the weight of each corresponding feature can be determined according to the contribution of each corresponding feature to the quality of the image to be evaluated; the weight of each corresponding feature and the corresponding features are processed through the fusion model based on the attention mechanism in the first model to determine the fusion feature.

[0172] In the disclosed embodiment, the feature vectors F1, F2, F3, F4, and F5 extracted from the above-mentioned different dimensions can be effectively fused to give play to the different roles of each feature. In the disclosed embodiment, a fusion strategy based on the attention mechanism is adopted. After adding the attention mechanism, the first model will automatically assign attention weights according to the contribution of different dimensional features to the quality of the image visual effect, which can better adapt to the characteristics of different images, improve the accuracy and reliability of the feature fusion effect, and make the fused feature vector more accurately reflect the quality of the image visual effect.

[0173] Step 203: Determine the quality evaluation value of the image to be evaluated based on the fusion features and the quality evaluation rules, and perform publishing processing on the image to be evaluated based on the quality evaluation value.

[0174] In the disclosed embodiment, after obtaining the fusion features, the first model can determine the quality assessment value of the image to be evaluated based on the fusion features and the quality assessment rules, and then publish the image to be evaluated based on the quality assessment value. Optionally, the quality assessment rules can include a correspondence between the fusion features and the quality assessment value, so that the quality assessment value of the image to be evaluated can be determined based on the fusion features and the quality assessment rules. Of course, the quality assessment rules can also be other content that can determine the quality assessment value, which is not limited in the disclosed embodiment.

[0175] As can be seen, in the disclosed embodiments, the images being evaluated are comprehensively assessed across multiple key dimensions, including portrait, saturation, composition, and content richness. Furthermore, the evaluation results from these dimensions, such as portrait, saturation, composition, and content richness, are organically integrated to establish a multi-dimensional comprehensive evaluation model for image visual quality, namely the first model. Furthermore, the interaction and weight distribution between factors in different dimensions are studied, thereby enabling accurate and comprehensive evaluation of image visual quality, significantly improving the accuracy of the evaluation of the images being evaluated.

[0176] In an embodiment of the present disclosure, after determining the quality evaluation value of the image to be evaluated, it can be determined whether to publish the image to be evaluated based on whether the quality evaluation value of the image to be evaluated is greater than a preset threshold. For example, if the quality evaluation value of the image to be evaluated is greater than the preset threshold, the image to be evaluated is published. For another example, if the quality evaluation value of the image to be evaluated is not greater than the preset threshold, the image to be evaluated is not published, for example, an evaluation analysis report of the image to be evaluated is fed back to the user so that it can be used to adjust the image to be evaluated, or to re-produce the image, that is, to provide guidance for the shooting, editing and optimization of the image, and to improve the efficiency and quality of image creation. In other words, the information processing method provided by the embodiment of the present disclosure can not only improve the accuracy of image quality evaluation, but also provide more powerful technical support for further optimization and application of images.

[0177] Exemplary devices

[0178] The exemplary embodiment of the present disclosure also provides an information processing device. Figure 9 As shown, the information processing device 900 includes the following program units:

[0179] A determination unit 901 is used to determine an image to be evaluated;

[0180] An extraction unit 902 is configured to perform multi-dimensional feature extraction processing on the image to be evaluated to obtain a fusion feature; the fusion feature is determined based on at least two of the object feature, saturation feature, composition feature, content fullness feature, and clarity feature;

[0181] The processing unit 903 is configured to determine a quality evaluation value of the image to be evaluated according to the fusion feature and the quality evaluation rule, so as to perform publishing processing on the image to be evaluated according to the quality evaluation value.

[0182] In a possible implementation, the extraction unit 902 is specifically configured to:

[0183] Performing feature extraction processing on the image to be evaluated using the first model to obtain corresponding features; wherein the feature extraction processing includes any one of object feature extraction processing, saturation feature extraction processing, composition feature extraction processing, content fullness feature extraction processing, and clarity feature extraction processing;

[0184] The corresponding features obtained are fused using the first model to obtain the fused features.

[0185] In a possible implementation, if the feature extraction process is an object feature extraction process, the extraction unit 902 is specifically configured to:

[0186] determining whether the image to be evaluated includes a portrait;

[0187] When it is determined that the image to be evaluated includes a portrait, performing portrait feature extraction processing on the image to be evaluated using the multi-task branch model in the first model to obtain multiple portrait sub-features; wherein the multiple portrait sub-features include at least two of the following: body sub-features, face sub-features, gender sub-features, clothing category sub-features, and face quality sub-features;

[0188] A portrait feature is determined based on the multiple portrait sub-features.

[0189] In a possible implementation, if the feature extraction process is a saturation feature extraction process, the extraction unit 902 is specifically configured to:

[0190] Performing color conversion on the image to be evaluated to obtain a converted first image;

[0191] The convolutional neural network model in the first model is used to perform feature extraction processing on the first image to obtain a saturation feature corresponding to the image to be evaluated.

[0192] In a possible implementation, if the feature extraction process is a composition feature extraction process, the extraction unit 902 is specifically configured to:

[0193] Performing global feature extraction processing on the image to be evaluated to obtain a first composition feature;

[0194] Performing local feature extraction processing on the image to be evaluated to obtain a second composition feature;

[0195] The composition feature is determined according to the first composition feature and the second composition feature.

[0196] In a possible implementation, the extraction unit 902 is specifically configured to:

[0197] Performing global feature extraction processing on the image to be evaluated through the transformer model in the first model to obtain a first composition feature.

[0198] In a possible implementation, the extraction unit 902 is specifically configured to:

[0199] Performing edge detection processing on the image to be evaluated to obtain edge detection information of the image to be evaluated;

[0200] The edge detection information is input into the convolutional neural network model in the first model to obtain a second composition feature.

[0201] In a possible implementation, if the feature extraction process is content richness feature extraction process, the extraction unit 902 is specifically configured to:

[0202] Determining the content and location information contained in the image to be evaluated through the multi-target detection and multi-label model in the first model;

[0203] Determine the content fullness feature of the image to be evaluated based on the content and position information.

[0204] In a possible implementation, if the feature extraction process is a clarity feature extraction process, the extraction unit 902 is specifically configured to:

[0205] Determining the clarity information corresponding to the image to be evaluated by using the physical feature extraction model in the first model;

[0206] Determine the clarity feature of the image to be evaluated based on the clarity information.

[0207] In a possible implementation, the extraction unit 902 is specifically configured to:

[0208] Determining the weight of each corresponding feature according to the contribution of each corresponding feature to the quality of the image to be evaluated;

[0209] The weights of the corresponding features and the corresponding features are processed through the fusion model based on the attention mechanism in the first model to determine the fusion feature.

[0210] In one possible implementation, the apparatus includes a training unit configured to:

[0211] Determining a plurality of images containing a plurality of features, and labeling each feature and a quality assessment value in the plurality of images containing the plurality of features to obtain training sample images;

[0212] Inputting the training sample image into a preset first model for training to obtain a plurality of prediction values, and comparing the prediction values ​​with the corresponding labeled values ​​in the training sample image to obtain a plurality of comparison results;

[0213] The preset first model is adjusted according to the multiple comparison results to obtain the first model.

[0214] In a possible implementation, the training unit is configured to:

[0215] Determining an overall loss function based on the multiple comparison results;

[0216] According to the overall loss function, the preset first model is adjusted to obtain the first model.

[0217] In a possible implementation, the training unit is configured to:

[0218] Determining first prediction values ​​of a plurality of portrait sub-features detected by the preset first model, and obtaining a first loss function according to a difference between the first prediction value and a first labeled value corresponding to the first prediction value;

[0219] Determining a second predicted value of a plurality of content fullness features detected by the preset first model, and obtaining a second loss function according to a difference between the second predicted value of the content fullness and a second labeled value corresponding to the second predicted value; and

[0220] Determining a quality assessment prediction value determined by the preset first model, and obtaining a third loss function according to a difference between the quality assessment prediction value and a quality assessment annotation value corresponding to the quality assessment prediction value;

[0221] The first loss function, the second loss function and the third loss function are weightedly calculated to obtain an overall loss function.

[0222] The specific details of each part of the above-mentioned device have been described in detail in the implementation method part. The undisclosed details can be found in the implementation method part, so they will not be repeated here.

[0223] It should be noted that although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the exemplary embodiments of the present disclosure, the features and functions of two or more modules or units described above can be concretized in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided into multiple modules or units to be concretized.

[0224] Exemplary Program Products

[0225] The exemplary embodiments of the present disclosure further provide a computer program product, which includes a computer program, and when the computer program is executed by a processor, implements the above-mentioned information processing method.

[0226] In one embodiment, a computer program product may be a tangible product containing a computer program, such as a computer-readable storage medium storing the computer program. The computer-readable storage medium may be a storage medium based on electrical, magnetic, optical, electromagnetic, infrared, or other signals, including but not limited to random access memory (RAM), read-only memory (ROM), magnetic tape, floppy disk, flash memory (Flash), mechanical hard disk drive (HDD), solid-state drive (SSD), and the like. Exemplarily, the computer program product may be implemented as a non-volatile storage medium storing the computer program, such as a read-only memory, a NAND flash memory (NandFlash), and the like.

[0227] In one embodiment, the computer program product may be an intangible product containing a computer program. For example, the computer program product may be implemented as a virtual digital product, such as a digital file such as an executable file or installation package storing the computer program.

[0228] The code of the computer program can be written in one or more programming languages. Programming languages ​​include C, Java, C++, etc. The program code can be executed entirely on the user computing device, partially on the user computing device, or as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device via any type of network, such as a local area network (LAN), a wide area network (WAN), etc., or can be connected to an external computing device (e.g., via an Internet connection provided by a carrier).

[0229] Computer programs can be carried or transmitted through electrical, magnetic, optical, electromagnetic, infrared and other signals. Electronic devices can convert signals carrying computer programs into digital signals, and then run the computer programs. When the computer program runs on an electronic device, its code is used to enable the electronic device to execute (more specifically, it can enable the processor of the electronic device to execute) the method steps of various exemplary embodiments of the present disclosure, such as the above-mentioned information processing method, which includes the following steps: Step 201: Determine the image to be evaluated; Step 202: Perform multi-dimensional feature extraction processing on the image to be evaluated to obtain fusion features; the fusion features are determined based on at least two of the object features, saturation features, composition features, content fullness features and clarity features; Step 203: Determine the quality assessment value of the image to be evaluated based on the fusion features and quality assessment rules, so as to publish the image to be evaluated based on the quality assessment value.

[0230] By implementing the above method steps through a computer program, the image to be evaluated can be first determined, and then multi-dimensional feature extraction can be performed on the image to be evaluated, such as object feature extraction, saturation feature extraction, composition feature extraction, and content fullness feature extraction, and then the extracted features can be fused to obtain fusion features, that is, fusion features determined based on at least two of the object features, saturation features, composition features, content fullness features, and clarity features. Furthermore, the quality assessment value of the image to be evaluated can be determined based on the fusion features and quality evaluation rules, so that the image to be evaluated can be published and processed based on the quality assessment value. It can be seen that in the embodiment of the present disclosure, the image to be evaluated can be comprehensively and carefully evaluated from multiple dimensions such as portrait, saturation, composition, and content fullness, as well as the interaction between multiple dimensions, to improve the accuracy of the evaluation.

[0231] Exemplary electronic devices

[0232] The exemplary embodiments of the present disclosure further provide an electronic device. The electronic device may include a processor and a memory. The memory stores executable instructions for the processor, such as a computer program. The processor executes the executable instructions to perform the method steps of various exemplary embodiments of the present disclosure.

[0233] Reference below Figure 10 , the electronic device is exemplarily described in the form of a general-purpose computing device. It should be understood that Figure 10 The electronic device 1000 shown is merely an example and should not limit the functions and scope of use of the embodiments of the present disclosure.

[0234] like Figure 10 As shown, the electronic device 1000 may include: a processor 1010 , a memory 1020 , a bus 1030 , an I / O (input / output) interface 1040 , and a network adapter 1050 .

[0235] The memory 1020 may include volatile memory, such as RAM 1021 and cache unit 1022, and may also include non-volatile memory, such as ROM 1023. The memory 1020 may also include one or more program modules 1024. Such program modules 1024 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. For example, the program modules 1024 may include the various units in the aforementioned apparatus.

[0236] The processor 1010 may include one or more processing units, for example: the processor 1010 may include an AP (Application Processor), a modem processor, a GPU (Graphics Processing Unit), an ISP (Image Signal Processor), a controller, an encoder, a decoder, a DSP (Digital Signal Processor), a baseband processor and / or an NPU (Neural-Network Processing Unit), etc.

[0237] The processor 1010 can be used to execute executable instructions stored in the memory 1020, such as the above-mentioned information processing method, which includes the following steps: Step 201: Determine the image to be evaluated; Step 202: Perform multi-dimensional feature extraction processing on the image to be evaluated to obtain fusion features; the fusion features are determined based on at least two of the object features, saturation features, composition features, content fullness features and clarity features; Step 203: Determine the quality evaluation value of the image to be evaluated based on the fusion features and quality evaluation rules, so as to publish the image to be evaluated based on the quality evaluation value.

[0238] By executing the above method steps through the processor 1010, the image to be evaluated can be first determined, and then multi-dimensional feature extraction can be performed on the image to be evaluated, such as object feature extraction, saturation feature extraction, composition feature extraction, and content fullness feature extraction, and then the extracted features can be fused to obtain a fusion feature, that is, a fusion feature determined based on at least two of the object features, saturation features, composition features, content fullness features, and clarity features. Furthermore, the quality assessment value of the image to be evaluated can be determined based on the fusion features and the quality assessment rules, so that the image to be evaluated can be published and processed based on the quality assessment value. It can be seen that in the embodiment of the present disclosure, a comprehensive and detailed assessment of the image to be evaluated can be performed from multiple dimensions such as portrait, saturation, composition, and content fullness, as well as the interaction between multiple dimensions, to improve the accuracy of the assessment.

[0239] The bus 1030 is used to realize the connection between different components of the electronic device 1000 and may include a data bus, an address bus, and a control bus.

[0240] The electronic device 1000 can communicate with one or more external devices 1100 (eg, a keyboard, a mouse, an external controller, etc.) through the I / O interface 1040 .

[0241] The electronic device 1000 can communicate with one or more networks via the network adapter 1050. For example, the network adapter 1050 can provide mobile communication solutions such as 3G / 4G / 5G / 6G, or wireless communication solutions such as wireless LAN, Bluetooth, and near-field communication. The network adapter 1050 can communicate with other modules of the electronic device 1000 via the bus 1030.

[0242] although Figure 10 Not shown, other hardware and / or software modules may also be provided in the electronic device 1000, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0243] As can be seen from the above, the technical solutions of the present disclosure can be implemented as methods, devices, systems, computer program products, storage media, electronic devices, etc. Those skilled in the art will understand that various aspects of the present disclosure can be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or an implementation that combines hardware and software aspects, such as "circuit", "module" or "system".

[0244] It should be understood that the present disclosure is not limited to the specific method steps or structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from the scope thereof. Those skilled in the art will easily think of other embodiments based on the specific embodiments provided by the present disclosure. Therefore, the specific embodiments provided by the present disclosure are merely exemplary, and the scope and spirit of the present disclosure are indicated by the claims, which should cover any variations, uses or adaptive changes of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary technical means in the field of the present technology that are not disclosed in the present disclosure.

Claims

1. An information processing method, characterized in that: The method comprises: Determine the image to be evaluated; Performing multi-dimensional feature extraction processing on the image to be evaluated to obtain a fusion feature; the fusion feature is determined based on at least two of the object feature, saturation feature, composition feature, content fullness feature, and clarity feature; The quality evaluation value of the image to be evaluated is determined according to the fusion features and the quality evaluation rule, so as to perform publishing processing on the image to be evaluated according to the quality evaluation value.

2. The method according to claim 1, characterized in that Performing multi-dimensional feature extraction processing on the image to be evaluated to obtain fusion features, including: Performing feature extraction processing on the image to be evaluated using the first model to obtain corresponding features; wherein the feature extraction processing includes any one of object feature extraction processing, saturation feature extraction processing, composition feature extraction processing, content fullness feature extraction processing, and clarity feature extraction processing; The corresponding features obtained are fused using the first model to obtain the fused features.

3. The method according to claim 2, characterized in that If the feature extraction process is an object feature extraction process, performing feature extraction on the image to be evaluated by using the first model to obtain corresponding features includes: determining whether the image to be evaluated includes a portrait; When it is determined that the image to be evaluated includes a portrait, performing portrait feature extraction processing on the image to be evaluated using the multi-task branch model in the first model to obtain multiple portrait sub-features; wherein the multiple portrait sub-features include at least two of the following: body sub-features, face sub-features, gender sub-features, clothing category sub-features, and face quality sub-features; A portrait feature is determined based on the multiple portrait sub-features.

4. The method according to claim 2, characterized in that If the feature extraction process is a saturation feature extraction process, performing feature extraction on the image to be evaluated by using the first model to obtain corresponding features includes: Performing color conversion on the image to be evaluated to obtain a converted first image; The convolutional neural network model in the first model is used to perform feature extraction processing on the first image to obtain a saturation feature corresponding to the image to be evaluated.

5. The method according to claim 2, characterized in that If the feature extraction process is a composition feature extraction process, performing feature extraction on the image to be evaluated by using the first model to obtain corresponding features includes: Performing global feature extraction processing on the image to be evaluated to obtain a first composition feature; Performing local feature extraction processing on the image to be evaluated to obtain a second composition feature; The composition feature is determined according to the first composition feature and the second composition feature.

6. The method according to claim 5, characterized in that Performing global feature extraction processing on the image to be evaluated to obtain a first composition feature includes: Performing global feature extraction processing on the image to be evaluated through the transformer model in the first model to obtain a first composition feature.

7. The method according to claim 5, characterized in that Performing local feature extraction processing on the image to be evaluated to obtain a second composition feature includes: Performing edge detection processing on the image to be evaluated to obtain edge detection information of the image to be evaluated; The edge detection information is input into the convolutional neural network model in the first model to obtain a second composition feature.

8. An information processing device, characterized in that The device comprises: a determination unit, configured to determine an image to be evaluated; an extraction unit, configured to perform multi-dimensional feature extraction processing on the image to be evaluated to obtain a fusion feature; the fusion feature is determined based on at least two of the object feature, saturation feature, composition feature, content fullness feature, and clarity feature; A processing unit is used to determine a quality evaluation value of the image to be evaluated according to the fusion feature and the quality evaluation rule, so as to perform publishing processing on the image to be evaluated according to the quality evaluation value.

9. An electronic device, characterized in that: include: processor; a memory for storing executable instructions of the processor; The processor is configured to perform the method according to any one of claims 1 to 7 by executing the executable instructions.

10. A computer program product having a computer program stored thereon, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images