Image quality evaluation method and system, electronic equipment and storage medium

Abstract

The embodiment of the invention provides an image quality evaluation method and system, electronic equipment and a storage medium. The method comprises the steps of obtaining a to-be-evaluated image; obtaining evaluation prompt information, wherein the evaluation prompt information comprises an evaluation dimension and a scoring range; the to-be-evaluated image and the evaluation prompt information are input into the multi-modal visual language model, single evaluation information and overall evaluation information of the to-be-evaluated image are obtained, each piece of single evaluation information corresponds to one evaluation dimension, and the single evaluation information comprises single scores of the evaluation dimensions and evaluation text information of the evaluation dimensions; the single scores are within a score range, the overall evaluation information corresponds to all evaluation dimensions, the overall evaluation information comprises grade information of the picture quality and summary text information of the picture quality, and the grade information is generated according to all the single scores; and combining the to-be-evaluated image, the evaluation prompt information, the single evaluation information and the overall evaluation information into an image quality evaluation report. According to the embodiment of the invention, personalized evaluation of the image quality is realized.

Description

Technical Field

[0001] The present application relates to the field of computer vision technology, and in particular to an image quality evaluation method, system, electronic device and storage medium. Background Art

[0002] Image quality assessment is a method to evaluate the quality of an image by analyzing and studying its characteristics. Image quality assessment plays a vital role in image processing, transmission, and storage. Effective image quality assessment helps improve the performance of image processing algorithms, improve the quality of images output by image processing algorithms, and meet the specific image quality requirements of different applications.

[0003] In the related art, the most widely used image quality assessment method is the image quality assessment method based on machine learning. This method inputs a large number of high-quality images and low-quality images into a deep learning model for model training, so that the deep learning model has the ability to predict image quality. However, in this image quality assessment method, the deep learning model uses the same evaluation dimensions for all input images, which are all dimensions specified during training and are relatively fixed, and cannot meet the user's personalized needs for image quality assessment. Summary of the invention

[0004] The present application provides an image quality evaluation method, system, electronic device and storage medium to solve the problem of poor personalization of image quality evaluation.

[0005] In a first aspect, the present application provides an image quality evaluation method, comprising:

[0006] Obtain the image to be evaluated;

[0007] Obtain evaluation prompt information, which includes evaluation dimensions and rating range;

[0008] Input the image to be evaluated and the evaluation prompt information into the pre-trained multimodal visual language model to obtain the single evaluation information and overall evaluation information of the image to be evaluated, where each single evaluation information corresponds to an evaluation dimension, and the single evaluation information includes a single score of the evaluation dimension and evaluation text information of the evaluation dimension. The single score is within the score range, and the overall evaluation information corresponds to all evaluation dimensions. The overall evaluation information includes the grade information of the picture quality and the summary text information of the picture quality, and the grade information is generated according to all the single scores;

[0009] The image to be evaluated, evaluation prompt information, single evaluation information and overall evaluation information are combined into an image quality evaluation report.

[0010] In some embodiments, obtaining evaluation prompt information includes:

[0011] Obtain input information on the input interface, which includes an option control and an input box control of the evaluation prompt template;

[0012] If the input information is selected information of an option control, the evaluation prompt template corresponding to the option control is determined as evaluation prompt information of the image to be evaluated;

[0013] If the input information is custom text information in the input box control, the custom text information is determined as evaluation prompt information of the image to be evaluated.

[0014] In some embodiments, obtaining evaluation prompt information includes:

[0015] Obtaining input information on an input interface, where the input interface includes an option control for evaluation dimensions and an option control for a rating range;

[0016] According to the input information including the selected information of the option control of the evaluation dimension and the selected information of the option control of the rating range, all the selected information is combined into the evaluation prompt information.

[0017] In some embodiments, obtaining evaluation prompt information includes:

[0018] Acquire input information on an input interface, the input interface including an input box control, a preview window and a confirmation control, the preview window having the same size as the image to be evaluated, the preview window being used to display the image to be evaluated, a calibration control being provided on the preview window, the calibration control being configured to move along a touch track in the preview window after being triggered, and the input information including first prompt information input in the input box control;

[0019] In response to a trigger instruction of the calibration control, determining a position of the calibration control according to an input touch track;

[0020] In response to the trigger instruction of the confirmation control, a second prompt information including an evaluation area is generated, where the evaluation area is the area determined by all the calibrated controls on the image to be evaluated.

[0021] Display the second prompt information in the input box control;

[0022] All text information in the input box control is determined as evaluation prompt information of the image to be evaluated.

[0023] In some embodiments, before combining the image to be evaluated, the evaluation prompt information, the single evaluation information and the overall evaluation information into an image quality evaluation report, the process further includes:

[0024] If the evaluation prompt information includes an evaluation area, a prompt box is added to the image to be evaluated, and the position of the prompt box is the same as the position of the evaluation area.

[0025] In some embodiments, when the image to be evaluated includes multiple images, the method further includes:

[0026] According to the overall evaluation information, the image quality evaluation reports of the plurality of images to be evaluated are combined into a first report set and a second report set, the first report set including the image quality evaluation reports whose grade information reaches a set grade, and the second report set including the image quality evaluation reports whose grade information does not reach the set grade;

[0027] According to the single evaluation information, the image quality evaluation reports in the second report set are combined into at least one third report set, and the third report set includes image quality evaluation reports in the second report set whose same single item scores are all lower than the set scores.

[0028] In a second aspect, the present application provides an image quality evaluation system, comprising:

[0029] An information input module is used to obtain the image to be evaluated and the evaluation prompt information;

[0030] A multimodal visual language model module, used for inputting the image to be evaluated and the evaluation prompt information into a pre-trained multimodal visual language model to obtain single evaluation information and overall evaluation information of the image to be evaluated, wherein the single evaluation information corresponds to the evaluation dimension in the evaluation prompt information, and the overall evaluation information corresponds to each single evaluation information;

[0031] The evaluation report generation module is used to combine the image to be evaluated, the evaluation prompt information, the single evaluation information and the overall evaluation information into an image quality evaluation report.

[0032] In some embodiments, the information input module further includes:

[0033] Information input interface, used to obtain input information.

[0034] In some embodiments, the information input interface includes:

[0035] Option controls and input box controls for the evaluation prompt template;

[0036] Option controls for evaluation dimensions and option controls for rating ranges.

[0037] In some embodiments, the information input module includes:

[0038] An image input module, used for acquiring the image to be evaluated;

[0039] The text input module is used to obtain the evaluation prompt information, and the evaluation prompt information includes evaluation dimensions.

[0040] In some embodiments, the image quality assessment system further includes: an output module for displaying the image quality assessment report.

[0041] In some embodiments, the image quality assessment system further includes: an output module for displaying an image quality assessment report.

[0042] In a third aspect, the present application provides an electronic device, which includes one or more processors and a memory, and also includes the system of the second aspect; the memory stores a computer program, the computer program includes computer instructions, and one or more processors call the computer instructions to enable the electronic device to execute the image quality assessment method of the first aspect.

[0043] In a fourth aspect, the present application provides a computer-readable storage medium, which includes a computer program. When the computer program is executed on an electronic device, the electronic device executes the image quality assessment method as in the first aspect.

[0044] It can be seen from the above technical content that the embodiment of the present application inputs the evaluation prompt information and the image to be evaluated into the multimodal visual language model together, and utilizes the multimodal visual language model's ability to process multimodal data to output the single evaluation information corresponding to the evaluation prompt information and the overall evaluation information that integrates all the single evaluation information, and combines the image to be evaluated, the evaluation prompt information, the single evaluation information and the overall evaluation information into an image quality evaluation report, so that users can view the personalized evaluation information of the image to be evaluated through a combination of pictures and texts, which improves user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solution of the present application, the drawings required for use in the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0046] Figure 1 A flowchart of an image quality evaluation method provided in an embodiment of the present application;

[0047] Figure 2 A schematic diagram of an image quality evaluation report provided in an embodiment of the present application;

[0048] Figure 3 A schematic diagram of another image quality evaluation report provided in an embodiment of the present application;

[0049] Figure 4 A schematic diagram of another image quality evaluation report provided in an embodiment of the present application;

[0050] Figure 5A schematic diagram of the structure of an image quality evaluation system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0051] The following embodiments are described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following embodiments do not represent all implementations consistent with the present application. They are only examples of systems and methods consistent with some aspects of the present application as detailed in the claims.

[0052] The embodiments of the present application provide an image quality assessment method, system, electronic device and storage medium. By pre-training a multimodal VLM (Visual Language Models) that supports both text and image input data, when performing image evaluation, personalized requirements for image evaluation can be input into the multimodal visual language model in the form of text together with the image to be evaluated, so as to obtain a personalized evaluation of the image and meet the user's personalized evaluation needs for the image.

[0053] See also Figure 1 , is a flow chart of an image quality evaluation method provided in an embodiment of the present application. Figure 1 As shown, the image quality evaluation method may include the following steps:

[0054] Step S100: obtaining an image to be evaluated.

[0055] In some embodiments, to facilitate image evaluation, a user may upload an image to be evaluated and input evaluation prompt information on an input interface of an image quality evaluation program, wherein the image quality evaluation program is configured to execute the image quality evaluation method described in an embodiment of the present application on an electronic device, and the electronic device may include a computer, a smart terminal, and other devices, and the evaluation prompt information represents the evaluation requirements for the image to be evaluated.

[0056] In some embodiments, the image formats supported by the image quality assessment program may include JPEG, PNG, BMP and other formats. Users can upload the image to be assessed in any of the above formats to the image quality assessment program through file upload or other input methods.

[0057] In some embodiments, multiple images to be evaluated may be uploaded on the input interface so as to perform image quality evaluation in batches.

[0058] Step S200: Obtain evaluation prompt information, which includes evaluation dimensions and scoring ranges.

[0059] In some embodiments, the input interface may be provided with an input box control, in which the user may input custom text information according to the evaluation requirements of the image to be evaluated. The custom text information may include evaluation requirements such as evaluation dimension, rating range, and evaluation area.

[0060] For example, if the user's evaluation requirement is to evaluate the clarity of an image, with a rating range of 1 to 5 points, the user can enter the following evaluation prompt information: "Please evaluate the clarity of the image, with a rating range of 1 to 5 points." The image quality evaluation program determines the user-defined text information entered in the input box control as the input evaluation prompt information.

[0061] For another example, if the user's evaluation requirement is to evaluate the clarity and color accuracy of an image, with a rating range of 1 to 5 points, the evaluation prompt information may be: "Please evaluate the clarity and color accuracy of the image, with a rating range of 1 to 5 points." The image quality evaluation program determines the user-defined text information entered in the input box control as the input evaluation prompt information.

[0062] In some embodiments, to facilitate the user to input the evaluation prompt information, the input box control may be provided with prompt information to prompt the user what style of evaluation prompt information to input. For example, the prompt information may be: "Please describe the evaluation dimension and rating range", "Please describe the image area that needs to be evaluated", etc.

[0063] In some embodiments, to facilitate users to quickly input evaluation prompt information, the input interface may be provided with option controls for multiple evaluation prompt templates, each option control corresponding to a type of evaluation prompt information. After the user selects one of the option controls, the image quality evaluation program determines the evaluation prompt information corresponding to the option control as the input evaluation prompt information.

[0064] Exemplarily, an evaluation prompt template is: "Please rate the image quality according to the six dimensions of "clarity, contrast, color accuracy, artifacts, noise, and exposure". The rating range is 1 to 5 points (1-very poor, 2-poor, 3-average, 4-good, 5-very good)." The evaluation prompt template includes evaluation dimensions and scoring criteria. The evaluation dimensions include six dimensions of clarity, contrast, color accuracy, artifacts, noise, and exposure, and the scoring criteria include situations with scores of 1 to 5.

[0065] In some embodiments, to facilitate users to input evaluation prompt information, the input interface may be provided with option controls for evaluation dimensions, such as clarity controls, contrast controls, color accuracy controls, artifact controls, noise controls, exposure controls, etc. These option controls correspond to different evaluation dimensions, respectively, and users can select one or more option controls for evaluation dimensions.

[0066] The input interface may also be provided with an option control for a scoring range, such as a control representing a scoring range of 1 to 3 points, a control representing a scoring range of 1 to 5 points, a control representing a scoring range of 1 to 10 points, and the user may select an option control for a scoring range.

[0067] After the user selects one or more option controls of the evaluation dimensions and selects an option control of a rating range, the image quality evaluation program can combine the information corresponding to the option controls selected by the user into evaluation prompt information of the image to be evaluated.

[0068] For example, if the user selects the following option controls: clarity control, contrast control, color accuracy control, artifact control, noise control, and controls representing a scoring range of 1 to 5 points, the image quality evaluation program can generate the following evaluation prompt information according to the set format: "Please rate the image quality according to the five dimensions of "clarity, contrast, color accuracy, artifacts, and noise" on a scale of 1 to 5 points (1-very poor, 2-poor, 3-average, 4-good, 5-very good)".

[0069] In some embodiments, to facilitate users to evaluate designated areas of an image to be evaluated rather than the entire area of ​​the image to be evaluated, after the user uploads the image to be evaluated, the input interface may display a preview window and a confirmation control. The preview window is the same size as the image to be evaluated and is used to display the complete image to be evaluated.

[0070] A calibration control is provided in the preview window, and the calibration control is used to limit the evaluation area. The calibration control is configured to move along the touch track in the preview window after being triggered.

[0071] For example, there are 4 calibration controls, specifically A, B, C, and D, which are located at the 4 top corners of the preview window by default, where A is located at the lower left corner, B is located at the lower right corner, C is located at the upper left corner, and D is located at the upper right corner. The user can trigger the calibration control by clicking the calibration control. After triggering the calibration control, drag the calibration control to update the position of the calibration control in the preview window. The calibration control follows the user's touch trajectory, such as the dragging trajectory, and updates its position in real time.

[0072] For another example, the number of calibration controls is 2, including a center point control O and a boundary point control P. The center point control O is located at the center of the preview window by default, and the boundary point control P is located on the boundary of the short side of the image to be evaluated by default. The area defined by the calibration controls is a circular area with the center point control O as the center and the distance between the center point control O and the boundary point control P as the radius. Both the center point control O and the boundary point control P are configured to move along the touch track in the preview window in response to being triggered, that is, the user can drag the center point control O to change its position, and can drag the boundary point control P to change its position.

[0073] After adjusting the positions of all calibration controls, the user can click the confirmation control corresponding to the preview window. The image quality evaluation program obtains the positions of the calibration controls in the preview window, determines the area enclosed by all calibration controls on the image to be evaluated as the evaluation area, and generates evaluation prompt information containing the evaluation area.

[0074] Example 1: After the user uploads an image to be evaluated, the positions of the four calibration controls A, B, C, and D are adjusted and the confirmation control is triggered, the image quality evaluation program can generate the following evaluation prompt information: "Please rate the image quality in the (a, b, c, d) area according to the six dimensions of "clarity, contrast, color accuracy, artifacts, noise, and exposure". The rating range is 1 to 5 points (1-very poor, 2-poor, 3-average, 4-good, 5-very good).", where a is the coordinate of A, b is the coordinate of B, c is the coordinate of C, and d is the coordinate of D.

[0075] Example 2: After the user uploads a 1920*1080 image to be evaluated, the positions of the two calibration controls O and P are adjusted and the confirmation control is triggered. The image quality evaluation program can generate the following evaluation prompt information: "Please rate the image quality in the (o, p) area according to the six dimensions of "clarity, contrast, color accuracy, artifacts, noise, and exposure". The rating range is 1 to 5 points (1-very poor, 2-poor, 3-average, 4-good, 5-very good).", where o1 is the coordinate of O and p is the coordinate of P.

[0076] By specifying an evaluation area in the preview window, the generated evaluation prompt information can include the evaluation area, so that the multimodal visual language model can evaluate the evaluation area, thereby realizing a customized evaluation area.

[0077] In some embodiments, the user may also specify the evaluation area of ​​the image to be evaluated in the custom text information. For example, the user may enter the following custom text information: "Please evaluate the image quality of the left half of the image." The "left half of the image" is the evaluation area.

[0078] In some embodiments, the input interface includes an input box control and the above-mentioned preview window and confirmation control. After the user enters the first prompt information in the input box control, the user can adjust the calibration control in the preview window, and then click the confirmation control. The image quality evaluation program will display the information generated according to the evaluation area as the second prompt information in the input box control. The user can edit the control displayed in the input box control again, for example, modify the coordinates of the evaluation area, delete or modify the repeated or contradictory information in the first prompt information and the second prompt information, etc. The input interface is provided with a submit control. After the user clicks the submit control, the image quality evaluation program determines all the text information in the input box control as the evaluation prompt information of the image to be evaluated.

[0079] In some embodiments, the image quality evaluation program is provided with an evaluation dimension comparison table, which includes the evaluation dimensions and aliases of the evaluation dimensions supported by the multimodal visual language model. After obtaining the custom text information input by the user, the custom text information can be segmented, and the segmented words representing the evaluation dimensions can be replaced with the evaluation dimensions in the evaluation dimension comparison table to improve the accuracy of the multimodal visual language model in identifying the evaluation dimensions.

[0080] If a word representing an evaluation dimension that is not included in the evaluation dimension comparison table is identified in the user-defined text information input by the user, a prompt message indicating that the evaluation dimension is not supported is generated and displayed, wherein the prompt message may also include evaluation dimensions included in the evaluation dimension comparison table and not included in the user-defined text information for the user to select. The user may choose to delete or modify the unsupported evaluation dimension, or not to process it, so that when the single evaluation information corresponding to the unsupported evaluation dimension is not included in the finally generated image quality evaluation report, the user can know the reason, thereby improving the user experience.

[0081] Step S300: Input the image to be evaluated and the evaluation prompt information into the pre-trained multimodal visual language model to obtain the single evaluation information and the overall evaluation information of the image to be evaluated, each single evaluation information corresponds to an evaluation dimension, the single evaluation information includes the single score of the evaluation dimension and the evaluation text information of the evaluation dimension, the single score is within the score range, the overall evaluation information corresponds to all evaluation dimensions, the overall evaluation information includes the grade information of the picture quality and the summary text information of the picture quality, and the grade information is generated based on all the single scores.

[0082] In some embodiments, after the user uploads the image to be evaluated and enters the evaluation prompt information on the input interface, the image quality evaluation program calls the pre-trained multimodal visual language model through the API interface, inputs the image to be evaluated and the evaluation prompt information into the multimodal visual language model, and obtains the output result of the multimodal visual language model.

[0083] The multimodal visual language model is a model that can learn from images and text. The multimodal visual language model is a generative model that can accept image and text input and generate text output. By training the open source multimodal visual language model with multimodal input data, the multimodal visual language model of the embodiment of the present application can be obtained.

[0084] The multimodal input data may include image data and text data. The text data may include exemplary evaluation prompt information and image evaluation information. The evaluation prompt information may be an evaluation prompt template, and different evaluation prompt templates correspond to different evaluation dimensions.

[0085] The evaluation prompt information may include descriptive information of the evaluation area, so that the multimodal visual language model has the ability to evaluate the image quality of the specified area. The descriptive information may be the coordinate information of the evaluation area, or it may be the area information specified by directional words such as "left half image", "right half image", etc. It may also be the descriptive information of a target object in the image, such as "people in the picture", "buildings in the image" and other information.

[0086] Image evaluation information may include single evaluation information and overall evaluation information. Single evaluation information is image evaluation information of a single evaluation dimension, such as clarity evaluation information, contrast evaluation information, brightness evaluation information, artifact evaluation information, noise evaluation information, color accuracy evaluation information, exposure evaluation information, resolution evaluation information, and other evaluation information. Single evaluation information includes single score and score description. If the evaluation prompt information includes a scoring standard, the multimodal visual language model outputs a single score and evaluation text information according to the scoring standard; if the evaluation prompt information does not include a scoring range, the multimodal visual language model outputs a single score and a scoring description according to the default scoring range.

[0087] The overall evaluation information corresponds to all evaluation dimensions, including the grade information of the picture quality and the summary text information of the picture quality. The grade information is generated based on all the individual scores, and the summary text information of the picture quality is the summary information of all the evaluation text information. Exemplarily, the grade information corresponds to the average score of all the individual scores, or corresponds to the weighted average score of all the individual scores. For example, the score range of the individual score is 1 to 5 points, 1-very poor, 2-poor, 3-average, 4-good, 5-very good. If the average score of all the individual scores is 4 points, the grade information is good. If the average score of all the individual scores is 3 points, the grade information is average.

[0088] In some embodiments, the rating information is also text information, that is, the overall evaluation information is all text information.

[0089] After the multimodal visual language model is trained, the image quality evaluation program can call the multimodal visual language model to process the image to be evaluated and the evaluation prompt information, and output the single evaluation information and overall evaluation information of the image to be evaluated, thereby realizing personalized evaluation of the image to be evaluated.

[0090] Step S400: combining the image to be evaluated, evaluation prompt information, single item evaluation information and overall evaluation information into an image quality evaluation report.

[0091] In some embodiments, after obtaining the single evaluation information and the overall evaluation information of the image to be evaluated, the image to be evaluated, the evaluation prompt information, the single evaluation information and the overall evaluation information can be combined into an image quality evaluation report. In this image quality evaluation report, the evaluation prompt information can be located above the image to be evaluated and on the same side, for example, on the left side of the report screen, and the single evaluation information and the overall evaluation information constitute the image quality evaluation information and are on the same side, for example, on the right side of the report screen. Figure 2 , Figure 3 and Figure 4 This combination of graphics and text makes it easier for users to view the image quality evaluation information of the image to be evaluated. Figure 2 For example, the image quality evaluation information includes overall evaluation information and individual evaluation information of six evaluation dimensions, namely, artifacts, noise, resolution, color, brightness and edge fidelity. Among them, the individual evaluation information of the evaluation dimension of artifacts includes the individual score as shown in the figure: Score: 1 (poor), and also includes the evaluation text information as shown in the figure: There are obvious light spot artifacts in the image... The overall evaluation information includes the grade information as shown in the figure: Overall, this is an image of average to poor quality. The overall evaluation information also includes the summary text information as shown in the figure: This image performs poorly in artifacts and resolution, the noise and color control are average, the brightness control is acceptable, but there is overexposure and the edge fidelity is poor.

[0092] In some embodiments, if the evaluation prompt information includes an evaluation area, a prompt box is added to the image to be evaluated, and the position of the prompt box is the same as the position of the evaluation area. The image to be evaluated with the prompt box added, the evaluation prompt information, the single evaluation information and the overall evaluation information are combined into an image quality evaluation report, so that the user can understand the evaluation area of ​​the image to be evaluated.

[0093] In some embodiments, after obtaining the single evaluation information and overall evaluation information of multiple images to be evaluated, a first report set and a second report set can be generated according to the overall evaluation information. The first report set includes image quality evaluation reports whose grade information reaches a set grade, and the set grade can be that the grade information is good or very good, for example, when the score range is 1 to 5 points, the average score of all single scores reaches 4 points or above, and the second report set includes image quality evaluation reports whose overall evaluation information does not reach the set grade, for example, when the score range is 1 to 5 points, the average score of all single scores does not reach 4 points. Through the first report set and the second report set, the batch uploaded images to be evaluated can be quickly divided into images with good image quality and images with poor image quality.

[0094] In some embodiments, for the second report set, a third report set can be generated based on the single evaluation information, and the third report set includes image quality evaluation reports in the second report set whose single score is lower than the set score. For example, when the score range is 1 to 5 points, the set score can be 4 points. The number of third report sets can be one or more. The third report set aggregates images with poor scores in a certain evaluation dimension, which is convenient for users to quickly view the problems of this batch of images in this evaluation dimension.

[0095] It can be seen from the above embodiments that the image quality evaluation method provided in the embodiments of the present application, by inputting the evaluation prompt information and the image to be evaluated into the multimodal visual language model together, and utilizing the multimodal visual language model's ability to process multimodal data, can output single evaluation information corresponding to the evaluation prompt information and overall evaluation information that integrates all single evaluation information, and combines the image to be evaluated, the evaluation prompt information, the single evaluation information and the overall evaluation information into an image quality evaluation report, so that users can conveniently view the personalized evaluation information of the image to be evaluated in a combination of pictures and texts, thereby improving the user experience.

[0096] The present application also provides an image quality evaluation system. Figure 5 The image quality evaluation system may include an information input module, a multimodal visual language model module, and an evaluation report generation module.

[0097] The information input module is used to obtain the image to be evaluated and the evaluation prompt information. In some embodiments, the information input module may include an information input interface, and the information input interface is used to obtain input information, that is, to obtain the image to be evaluated and the evaluation prompt information.

[0098] In some embodiments, the information input module may include an option control and an input box control for an evaluation prompt template, as well as an option control for evaluation dimensions and an option control for a rating range.

[0099] In some embodiments, the information input module may include an image input module and a text input module, the image input module is used to obtain the image to be evaluated, and the text input module is used to obtain evaluation prompt information, and the evaluation prompt information includes evaluation dimensions.

[0100] The multimodal visual language model module is used to input the image to be evaluated and the evaluation prompt information into the pre-trained multimodal visual language model to obtain the single evaluation information and the overall evaluation information of the image to be evaluated. The single evaluation information corresponds to the evaluation dimension in the evaluation prompt information, and the overall evaluation information corresponds to each single evaluation information.

[0101] The evaluation report generation module is used to combine the image to be evaluated, evaluation prompt information, single evaluation information and overall evaluation information into an image quality evaluation report.

[0102] In some embodiments, the image quality assessment system further includes an output module, and the output module is used to display an image quality assessment report.

[0103] In some embodiments, the output module is further used to download the image quality assessment report.

[0104] In some embodiments, the working process and technical effects of each module of the image quality evaluation system can refer to the process description in the image quality evaluation method, which will not be repeated here.

[0105] An embodiment of the present application also provides an electronic device, which includes one or more processors and a memory, and also includes the system of the second aspect; the memory stores a computer program, the computer program includes computer instructions, and one or more processors call the computer instructions to enable the electronic device to execute the image quality evaluation method of the embodiment of the present application.

[0106] An embodiment of the present application further provides a computer-readable storage medium, which includes a computer program. When the computer program runs on an electronic device, the electronic device executes the image quality evaluation method of the embodiment of the present application.

[0107] Similar parts between the embodiments provided in this application can be referenced to each other. The specific implementation methods provided above are only a few examples under the general concept of this application and do not constitute a limitation on the protection scope of this application. For those skilled in the art, any other implementation methods expanded based on the scheme of this application without creative work belong to the protection scope of this application.

Claims

1. A method for evaluating image quality, characterized in that: include: Obtain the image to be evaluated; Obtaining evaluation prompt information, wherein the evaluation prompt information includes evaluation dimensions and scoring range; Inputting the image to be evaluated and the evaluation prompt information into a pre-trained multimodal visual language model to obtain single evaluation information and overall evaluation information of the image to be evaluated, wherein each single evaluation information corresponds to one evaluation dimension, the single evaluation information includes a single score of the evaluation dimension and evaluation text information of the evaluation dimension, the single score is within the score range, the overall evaluation information corresponds to all the evaluation dimensions, the overall evaluation information includes grade information of picture quality and summary text information of the picture quality, and the grade information is generated according to all the single scores; The image to be evaluated, evaluation prompt information, single item evaluation information and overall evaluation information are combined into an image quality evaluation report.

2. The image quality evaluation method according to claim 1, characterized in that: The obtaining of evaluation prompt information includes: Acquire input information on an input interface, wherein the input interface includes an option control and an input box control of an evaluation prompt template; If the input information is selected information of the option control, determining the evaluation prompt template corresponding to the option control as the evaluation prompt information of the image to be evaluated; If the input information is user-defined text information in the input box control, the user-defined text information is determined as evaluation prompt information of the image to be evaluated.

3. The image quality evaluation method according to claim 1, characterized in that: The obtaining of evaluation prompt information includes: Acquiring input information on an input interface, wherein the input interface includes an option control for an evaluation dimension and an option control for a rating range; According to the input information including the selected information of the option control of the evaluation dimension and the selected information of the option control of the rating range, all the selected information is combined into the evaluation prompt information.

4. The image quality evaluation method according to claim 1, characterized in that: The obtaining of evaluation prompt information includes: Acquire input information on an input interface, the input interface comprising an input box control, a preview window and a confirmation control, the preview window having the same size as the image to be evaluated, the preview window being used to display the image to be evaluated, a calibration control being provided on the preview window, the calibration control being configured to move along a touch track in the preview window after being triggered, and the input information comprising first prompt information input in the input box control; In response to a trigger instruction of the calibration control, determining a position of the calibration control according to an input touch track; In response to the trigger instruction of the confirmation control, a second prompt information including an evaluation area is generated, where the evaluation area is an area determined by all the calibration controls on the image to be evaluated, Displaying the second prompt information in the input box control; All text information in the input box control is determined as evaluation prompt information of the image to be evaluated.

5. The image quality evaluation method according to claim 4, characterized in that: Before combining the image to be evaluated, the evaluation prompt information, the single evaluation information and the overall evaluation information into an image quality evaluation report, the method further includes: If the evaluation prompt information includes an evaluation area, a prompt box is added to the image to be evaluated, and the position of the prompt box is the same as the position of the evaluation area.

6. The image quality evaluation method according to claim 1, characterized in that: When the to-be-evaluated images include a plurality of images, the method further includes: According to the overall evaluation information, the image quality evaluation reports of the plurality of images to be evaluated are combined into a first report set and a second report set, wherein the first report set includes the image quality evaluation reports whose grade information reaches a set grade, and the second report set includes the image quality evaluation reports whose grade information does not reach the set grade; According to the single evaluation information, the image quality evaluation reports in the second report set are combined into at least one third report set, and the third report set includes image quality evaluation reports in the second report set whose single scores are all lower than the set scores.

7. An image quality evaluation system, characterized in that: include: An information input module is used to obtain the image to be evaluated and the evaluation prompt information; A multimodal visual language model module, used for inputting the image to be evaluated and the evaluation prompt information into a pre-trained multimodal visual language model to obtain single evaluation information and overall evaluation information of the image to be evaluated, wherein the single evaluation information corresponds to the evaluation dimension in the evaluation prompt information, and the overall evaluation information corresponds to each single evaluation information; The evaluation report generation module is used to combine the image to be evaluated, the evaluation prompt information, the single evaluation information and the overall evaluation information into an image quality evaluation report.

8. The image quality evaluation system according to claim 7, characterized in that: The information input module also includes: Information input interface, used to obtain input information.

9. The image quality evaluation system according to claim 8, characterized in that: The information input interface includes: Option controls and input box controls for the evaluation prompt template; Option controls for evaluation dimensions and option controls for rating ranges.

10. The image quality evaluation system according to claim 7, characterized in that: The information input module comprises: An image input module, used for acquiring the image to be evaluated; The text input module is used to obtain the evaluation prompt information, and the evaluation prompt information includes evaluation dimensions.

11. The image quality evaluation system according to claim 7, characterized in that: The image quality evaluation system further includes: an output module, which is used to display the image quality evaluation report.

12. An electronic device comprising one or more processors and a memory, characterized in that: It also includes the system described in any one of claims 7-11; the memory stores a computer program, the computer program includes computer instructions, and the one or more processors call the computer instructions to enable the electronic device to execute the image quality assessment method described in any one of claims 1-6.

13. A computer-readable storage medium, characterized in that: It comprises a computer program, and when the computer program is run on an electronic device, the electronic device executes the image quality assessment method as described in any one of claims 1 to 6.

Images