Image rendering quality detection method and device, computer device, medium and product

Abstract

The application relates to an image rendering quality detection method and device, computer equipment, a storage medium and a computer program product. The application comprises the following steps: obtaining a standard picture frame presented by a standard software rendering a to-be-detected video; determining a current picture frame range corresponding to a current behavior event to which a current detection picture frame belongs, when the to-be-detected software renders the to-be-detected video; obtaining current frame detection semantic information of the current detection picture frame; and predicting detection semantic information of picture frames after the current detection picture frame in the current picture frame range. According to the current frame detection semantic information and the predicted detection semantic information, and corresponding standard semantic information, a similarity between the current detection picture frame and a current standard picture frame is obtained, and the rendering distortion degree of the to-be-detected software is determined according to the similarity. The application can automatically identify semantic information in the to-be-detected video and detect the rendering quality of the to-be-detected software, thereby improving the detection efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of graphics processor testing, and particularly relates to an image rendering quality detection method and device, a computer device, a storage medium and a computer program product. BACKGROUND

[0002] The graphics rendering quality of a graphics processing unit (GPU) is one of important parameters for evaluating the performance of the GPU, and the graphics rendering quality can be analyzed through the display effect of an image. Specifically, the display effect of the image is divided into three categories: severe distortion, slight distortion and no distortion.

[0003] In a traditional method, a special test software is usually used to detect the distortion of an image or a video to be detected. However, the test software needs to be pre-installed locally on a terminal for detection, and in a scenario where multiple terminals are detected simultaneously, the detection time is longer, which reduces the detection efficiency. SUMMARY

[0004] Therefore, it is necessary to provide an image rendering quality detection method and device, a computer device, a computer readable storage medium and a computer program product, which can improve the detection efficiency of the image rendering quality.

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

[0006] obtaining a behavior event axis of a video to be detected, the behavior event axis being used to indicate at least one behavior event involved in the video to be detected and a corresponding picture frame range of each behavior event;

[0007] determining a corresponding current picture frame range of a current behavior event to which a current detection picture frame presented when a software to be detected renders the video to be detected belongs;

[0008] obtaining current frame detection semantic information of the current detection picture frame, and predicting detection semantic information of picture frames after the current detection picture frame in the current picture frame range according to the current frame detection semantic information;

[0009] obtaining a similarity between the current detection picture frame and a standard picture frame according to the current frame detection semantic information and the predicted detection semantic information, and current frame standard semantic information of the standard picture frame and standard semantic information of picture frames after the standard picture frame in the current picture frame range, the standard picture frame being presented by a standard software rendering the video to be detected;

[0010] determining a rendering distortion degree of the software to be detected according to the similarity between the current detection picture frame and the standard picture frame.

[0011] In one of the embodiments, the method further comprises:

[0012] After obtaining the standard semantic information of the standard picture frame, obtain the super setting information;

[0013] Based on the super setting information, check the standard semantic information, and in case of failure, modify the standard semantic information that fails to pass the check based on the super setting information.

[0014] In one of the embodiments, the step of obtaining the similarity between the current detection picture frame and the current standard picture frame according to the current frame detection semantic information and the predicted detection semantic information, and the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame within the current picture frame range, comprises:

[0015] Integrate the current frame detection semantic information and the predicted detection semantic information to obtain the current detection semantic information;

[0016] Integrate the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame within the current picture frame range to obtain the current standard semantic information;

[0017] Obtain the similarity between the current detection semantic information and the current standard semantic information as the similarity between the current detection picture frame and the current standard picture frame.

[0018] In one of the embodiments, the step of obtaining the similarity between the current detection picture frame and the current standard picture frame according to the current frame detection semantic information and the predicted detection semantic information, and the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame within the current picture frame range, comprises:

[0019] Within the current picture frame range, obtain a picture frame comparison range determined by the current detection picture frame and the detection picture frame after the current detection picture frame;

[0020] For each pair of detection picture frame and standard picture frame with consistent frame sequence within the picture frame comparison range, obtain the similarity between the detection semantic information of the detection picture frame and the standard semantic information of the standard picture frame in each pair of detection picture frame and standard picture frame;

[0021] According to the similarity corresponding to each pair of detection picture frame and standard picture frame within the picture frame comparison range, determine the similarity between the current detection picture frame and the current standard picture frame.

[0022] In one of the embodiments, the process of identifying semantic information is implemented by a semantic analysis model, and the process of predicting semantic information is implemented by a semantic prediction model; the method further comprises:

[0023] obtaining a distortion degree of each detection picture frame presented when the to-be-detected software draws the to-be-detected video, the distortion degree of each detection picture frame being determined according to a similarity between each detection picture frame and a corresponding standard picture frame;

[0024] generating a detection report for reflecting drawing quality of the to-be-detected software according to the detection picture frame with the distortion degree greater than the preset degree and the corresponding distortion degree, and generating a respective model analysis log according to a respective model running record of the semantic analysis model and the semantic prediction model.

[0025] In one of the embodiments, the step of obtaining the current-frame detection semantic information of the current detection picture frame comprises:

[0026] identifying the current detection picture frame by the semantic analysis model to obtain identified text;

[0027] performing word segmentation on the identified text, and extracting keywords from the word segmentation result to obtain the current-frame detection semantic information.

[0028] In a second aspect, the application further provides an image drawing quality detection device, comprising:

[0029] a behavior acquisition module configured to acquire a behavior event axis of a to-be-detected video, the behavior event axis being used to indicate at least one behavior event involved in the to-be-detected video and a corresponding picture frame range of each behavior event;

[0030] a range determination module configured to determine, for a current detection picture frame presented when a to-be-detected software draws a to-be-detected video, a current picture frame range of a current behavior event to which the current detection picture frame belongs;

[0031] a semantic prediction module configured to obtain current-frame detection semantic information of the current detection picture frame, and predict detection semantic information of picture frames after the current detection picture frame within the current picture frame range according to the current-frame detection semantic information;

[0032] a similarity acquisition module configured to acquire a similarity between the current detection picture frame and a current standard picture frame according to the current-frame detection semantic information and the predicted detection semantic information, and according to current-frame standard semantic information of the current standard picture frame and standard semantic information of picture frames after the current standard picture frame within the current picture frame range, the current standard picture frame being presented by a standard software drawing the to-be-detected video;

[0033] The distortion determination module is configured to determine a drawing distortion degree of the to-be-detected software according to the similarity between the current detection picture frame and the current standard picture frame.

[0034] In a third aspect, the present application provides a computer device, comprising a memory and a processor, the memory stores a computer program, and the processor implements the method steps of any one of the first aspect when executing the computer program.

[0035] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the method steps of any one of the first aspect.

[0036] In a fifth aspect, the present application provides a computer program product, comprising a computer program, and the computer program is executed by a processor to implement the method steps of any one of the first aspect.

[0037] The image drawing quality detection method, device, computer device, storage medium and computer program product can automatically identify semantic information in the to-be-detected video, and detect the drawing quality of the to-be-detected software, thereby improving the detection efficiency of the image drawing quality. BRIEF DESCRIPTION OF DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the embodiments or the related art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0039] Figure 1 It is an application environment diagram of the image drawing quality detection method in one embodiment;

[0040] Figure 2 It is a flowchart of the image drawing quality detection method in one embodiment;

[0041] Figure 3 a flowchart of the process of obtaining the similarity in an embodiment;

[0042] Figure 4 a flowchart of the process of obtaining the similarity in an embodiment;

[0043] Figure 5 a flowchart of the image rendering quality detection method in an embodiment;

[0044] Figure 6 a composition diagram of the behavior event axis in an embodiment;

[0045] Figure 7 a structural block diagram of the image rendering quality detection device in an embodiment;

[0046] Figure 8 an internal structure diagram of the computer device in an embodiment. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application.

[0048] The image rendering quality detection method provided by the embodiments of the present application can be applied to, for example, Figure 1The application environment is shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data required by the server 104 to process. The data storage system can be integrated on the server 104, or placed on the cloud or other network servers. Among them, the terminal 102 is used to obtain the behavior event axis of the to-be-detected video through the server 104, the behavior event axis is used to indicate at least one behavior event involved in the to-be-detected video and the corresponding picture frame range of each behavior event; for the current detection picture frame presented by the to-be-detected software when drawing the to-be-detected video, determine the current picture frame range corresponding to the current behavior event to which the current detection picture frame belongs; obtain the current frame detection semantic information of the current detection picture frame, according to the current frame detection semantic information, predict the detection semantic information of the picture frame after the current detection picture frame in the current picture frame range; according to the current frame detection semantic information and the predicted detection semantic information, and the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame in the current picture frame range, obtain the similarity between the current detection picture frame and the current standard picture frame, the standard picture frame is presented by the standard software when drawing the to-be-detected video; according to the similarity between the current detection picture frame and the current standard picture frame, determine the distortion degree of the to-be-detected software. Among them, the terminal 102 can be, but not limited to, various personal computers, notebook computers and the like. The server 104 can be realized by an independent server or a server cluster composed of multiple servers.

[0049] In an exemplary embodiment, as shown in Figure 2 , a picture drawing quality detection method is provided. The method is applied to the terminal 102 in Figure 1 for example, including the following steps 202 to 210. Among them:

[0050] S202: Obtain the behavior event axis of the to-be-detected video, and the behavior event axis is used to indicate at least one behavior event involved in the to-be-detected video and the corresponding picture frame range of each behavior event.

[0051] The behavior event axis refers to an event axis obtained by labeling all behaviors on a time axis by identifying various behaviors in the video to be detected, and the behavior event refers to behavior information corresponding to each picture frame in the video to be detected. The picture frame range refers to the number of picture frames covered by each behavior event. The picture frame refers to a picture presented after the video to be detected is rendered by the GPU software. If there is an error in the software when rendering each video frame in the video to be detected, the picture presented after rendering may have an image distortion problem. Specifically, the image distortion usually includes color distortion and texture distortion. The color distortion refers to the inconsistency between the picture presented after rendering and the common sense or standard image. The texture distortion refers to the tearing, blurring, deformation, mosaic and other problems of the picture presented after rendering.

[0052] In the traditional method, when judging whether the picture frame has an image distortion problem, the detection personnel usually needs to observe each picture frame in the video to be detected to judge whether the color in the picture frame is normal and whether the texture is normal. This requires the detection personnel to continuously observe the video to be detected for a long time, and the detection efficiency is low. Therefore, in the embodiments of the present application, a picture rendering quality detection method is provided. The video to be detected is rendered by standard software to obtain the behavior event axis of the video to be detected. The behavior event axis is used to indicate at least one behavior event involved in the video to be detected and the picture frame range corresponding to each behavior event, as standard information, and compared with the actual information obtained when the video to be detected is rendered by the detection software to realize automatic detection of the detection software.

[0053] Specifically, the behavior information is obtained by identifying the picture frame by the semantic analysis model. It can be understood that the behavior of a person or other object is identified and understood from the picture frame, and the behavior is described as a natural language sentence, such as identifying the behavior of a person "opening the door" from the picture frame, and then converting it into a semantic description such as "a person opens the door".

[0054] S204: For the current detection picture frame presented when the video to be detected is rendered by the detection software, determine the current picture frame range corresponding to the current behavior event to which the current detection picture frame belongs.

[0055] Among them, the software to be detected refers to the GPU software to be detected. When the video to be detected is drawn by the software to be detected, the terminal locates the current detection picture frame in the behavior event axis according to the frame sequence corresponding to the current detection picture frame presented by the software to be detected, and determines the current behavior event to which the current detection picture frame belongs, as well as the current picture frame range corresponding to the current behavior event. Specifically, the frame sequence corresponding to the current detection picture frame can be determined by the current input duration of the video to be detected in the software to be detected, or each frame of the video to be detected can be pre-marked with a serial number to determine the corresponding position of the current detection picture frame on the behavior event axis.

[0056] S206: Acquire current frame detection semantic information of the current detection frame, and predict detection semantic information of frames following the current detection frame within the current frame range based on the current frame detection semantic information.

[0057] The terminal identifies the current frame detection semantic information of the current detection frame through a language analysis model. Specifically, the language analysis model can be a natural language process (NLP) model. The NLP model processes the current detection frame, extracts features of the current detection frame, and generates a description statement for the current detection frame based on the features to obtain the current frame detection semantic information. For the frames following the current detection frame within the current frame range, the detection semantic information corresponding to each frame following the current detection frame is predicted based on the current frame detection semantic information.

[0058] Specifically, the terminal uses a language prediction model to obtain semantic information about the frames following the current one. This language prediction model can be a multimodal fusion anthropomorphic model. Multimodal fusion integrates and processes information from different sensory modalities (such as vision, speech, and language) to achieve better human-computer interaction. Multimodal fusion improves information volume and comprehension. The information provided by different sensory modalities complements each other, resulting in higher information volume and stronger comprehension. Anthropomorphic models are a type of artificial intelligence model that can engage in natural conversations with people through voice, text, and emojis, demonstrating human-like communication skills. Through big data learning, they continuously enrich their knowledge base and response capabilities, approaching human intelligence levels. Examples include virtual assistants, intelligent robots, and online customer service.

[0059] Further, in training the multi-modal fusion personification model, mainly through a deep learning network, the deep learning network can extract feature information of each picture frame, integrate the feature information of each picture frame before and after the picture frame through the model, form a series of complete action features, and divide different actions, so that each window in the model covers one action. Through the learning and training of big data, when using the multi-modal fusion personification model to predict the detection semantic information of the picture frame after the current detection picture frame in the current picture frame range, the action feature described by the current frame detection semantic information is determined to belong to the action category, and then the corresponding detection semantic information of each picture frame after is predicted.

[0060] S208: According to the current frame detection semantic information and the predicted detection semantic information, and the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame in the current picture frame range, the similarity between the current detection picture frame and the current standard picture frame is obtained. The standard picture frame is drawn by a standard software to present the video to be detected.

[0061] Among them, by comparing the current frame detection semantic information and the predicted detection semantic information with the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame in the current picture frame range, the similarity between the current detection picture frame and the current standard picture frame is calculated. The similarity represents the correctness of the current detection picture frame drawn. In the case that the software to be detected does not appear abnormal, the similarity between the current detection picture frame and the current standard picture frame should be close to 1. In the case that the software to be detected appears abnormal and causes the presented picture frame to appear image distortion, the similarity will be less than 1. Among them, the standard picture frame is drawn by a standard software to present the video to be detected, and the frame standard semantic information is obtained by recognizing the standard picture frame through a language analysis model.

[0062] S210: According to the similarity between the current detection picture frame and the current standard picture frame, the drawing distortion degree of the software to be detected is determined.

[0063] The higher the similarity between the current detection picture frame and the current standard picture frame, the lower the drawing distortion degree of the software to be detected, wherein the drawing distortion degree includes severe distortion, slight distortion, and no distortion. Specifically, the severe distortion refers to problems such as blurring, collapsed block, mosaic, color block, fault, screen, sawtooth, artifact, color shift, and the like of the current detection picture frame. The slight distortion refers to problems such as detail blurring, color tone shift, obvious sawtooth, slightly obvious residual image, small color block, not obvious fault, few screen speckles, and fine mosaic of the current detection picture frame. By setting different similarity thresholds, the drawing distortion degree of the software to be detected can be divided. Further, in the case that the software to be detected has image distortion, the current detection picture frame with image distortion and the similarity corresponding to the picture frame can be saved in a specified file to generate a detection report reflecting the drawing quality of the software to be detected.

[0064] In the above image drawing quality detection method, by obtaining the standard picture frame presented by the standard software drawing the video to be detected, the current detection picture frame presented by the software to be detected drawing the video to be detected is determined to belong to the current picture frame range corresponding to the current behavior event, the current frame detection semantic information of the current detection picture frame is obtained, and the detection semantic information of the picture frame after the current detection picture frame in the current picture frame range is predicted according to the current frame detection semantic information. The similarity between the current detection picture frame and the current standard picture frame is obtained according to the current frame detection semantic information and the predicted detection semantic information, and the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame in the current picture frame range. The drawing distortion degree of the software to be detected is determined according to the similarity, which can automatically identify the semantic information in the video to be detected and detect the drawing quality of the software to be detected, thereby improving the detection efficiency of the image drawing quality.

[0065] In an exemplary embodiment, the method further comprises: after obtaining the standard semantic information of the standard picture frame, obtaining the super setting information; verifying the standard semantic information based on the super setting information, and modifying the standard semantic information that fails the verification based on the super setting information in the case that the verification fails.

[0066] In the process of drawing the to-be-detected video by the standard software, in order to eliminate the interference caused by the color distortion of the to-be-detected video itself, after the standard semantic information of the standard picture frame is obtained, the standard semantic information needs to be checked. Specifically, by setting the super setting information, the standard semantic information that fails the check is modified. For example, if the standard semantic information of a standard picture frame includes a green-skinned humanoid monster, it indicates that the to-be-detected video is abnormal, and the corresponding super setting information can be that the person becomes a monster. In this way, when the language analysis model identifies the standard semantic information, a mapping relationship between the monster and the person is established, so that the modification of the standard semantic information is realized.

[0067] In the embodiment, after the standard semantic information of the standard picture frame is obtained, the super setting information is obtained, the standard semantic information is checked based on the super setting information, and in the case that the check fails, the standard semantic information that fails the check is modified based on the super setting information. The influence of the distortion of the to-be-detected video on the detection of the drawing quality of the to-be-detected software can be eliminated, so that the detection accuracy of the image drawing quality is improved.

[0068] In one exemplary embodiment, as shown in Figure 3 The step of obtaining the similarity between the current detection picture frame and the current standard picture frame includes the following steps 302 to 306. Wherein:

[0069] S302: The current frame detection semantic information and the predicted detection semantic information are integrated to obtain the current detection semantic information.

[0070] In the process of obtaining the similarity between the current detection picture frame and the current standard picture frame, the current frame detection semantic information and the predicted detection semantic information can be integrated, for example, the union of the current frame detection semantic information and the predicted detection semantic information is taken, and finally a current detection semantic information representing the current detection picture frame is obtained.

[0071] S304: The current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame within the current picture frame range are integrated to obtain the current standard semantic information.

[0072] In the process of obtaining the similarity between the current detection picture frame and the current standard picture frame, the current frame detection semantic information and the predicted detection semantic information can be integrated, for example, the union of the current frame detection semantic information and the predicted detection semantic information is taken, and finally a current detection semantic information representing the current detection picture frame is obtained.

[0073] S306: Obtain the similarity between the current detection semantic information and the current standard semantic information as the similarity between the current detection picture frame and the current standard picture frame.

[0074] In the embodiment, the current detection semantic information and the predicted detection semantic information are integrated to obtain the current detection semantic information, the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame in the current picture frame range are integrated to obtain the current standard semantic information, and the similarity between the current detection semantic information and the current standard semantic information is obtained as the similarity between the current detection picture frame and the current standard picture frame, so that the accuracy of the similarity can be ensured, and the accurate detection of the rendering quality of the to-be-detected software can be realized.

[0075] In the embodiment, the current detection semantic information and the predicted detection semantic information are integrated to obtain the current detection semantic information, the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame in the current picture frame range are integrated to obtain the current standard semantic information, and the similarity between the current detection semantic information and the current standard semantic information is obtained as the similarity between the current detection picture frame and the current standard picture frame, so that the accuracy of the similarity can be ensured, and the accurate detection of the rendering quality of the to-be-detected software can be realized.

[0076] In one exemplary embodiment, as shown in Figure 4 According to the current frame detection semantic information and the predicted detection semantic information, and the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame in the current picture frame range, the similarity between the current detection picture frame and the current standard picture frame is obtained, including the following steps 402 to step 406. Wherein:

[0077] S402: In the current picture frame range, a picture frame comparison range determined by the current detection picture frame and the detection picture frame after the current detection picture frame is obtained.

[0078] In the embodiment, the current detection picture frame and all detection picture frames after the current detection picture frame can be constructed into the picture frame comparison range, and each detection picture frame in the picture frame comparison range is compared with the corresponding standard picture frame in sequence.

[0079] S404: For each pair of detection picture frame and standard picture frame with consistent frame sequence in the picture frame comparison range, the similarity between the detection semantic information of the detection picture frame and the standard semantic information of the standard picture frame in each pair of detection picture frame and standard picture frame is obtained.

[0080] The similarity of each pair of detection picture frame and standard picture frame in the picture frame comparison range is determined according to the similarity corresponding to each pair of detection picture frame and standard picture frame in the picture frame comparison range.

[0081] The similarity between the current detection picture frame and the current standard picture frame is determined according to the similarity corresponding to each pair of detection picture frame and standard picture frame in the picture frame comparison range.

[0082] The similarity between the current detection picture frame and the current standard picture frame is determined according to the similarity corresponding to each pair of detection picture frame and standard picture frame in the picture frame comparison range.

[0083] In the embodiment, the picture frame comparison range determined by the current detection picture frame and the detection picture frame after the current detection picture frame is obtained in the current picture frame range, the similarity between the detection semantic information of the detection picture frame and the standard semantic information of the standard picture frame in each pair of detection picture frame and standard picture frame in the picture frame comparison range is obtained, the similarity between the current detection picture frame and the current standard picture frame is determined according to the similarity corresponding to each pair of detection picture frame and standard picture frame in the picture frame comparison range, and the accuracy of the similarity can be ensured, so that the drawing quality of the to-be-detected software can be accurately detected.

[0084] In one exemplary embodiment, the process of identifying semantic information is realized by a semantic analysis model, and the process of predicting semantic information is realized by a semantic prediction model; the method further includes: obtaining the distortion degree of each detection picture frame presented when the to-be-detected software draws the to-be-detected video, the distortion degree of each detection picture frame being determined according to the similarity between each detection picture frame and the corresponding standard picture frame; generating a detection report reflecting the drawing quality of the to-be-detected software according to the detection picture frame with a distortion degree greater than a preset degree and the corresponding distortion degree; and generating a respective model analysis log according to the model running record of each of the semantic analysis model and the semantic prediction model.

[0085] Wherein, for each detection picture frame, the higher the similarity between the detection picture frame and the corresponding standard picture frame, the lower the distortion degree of the detection picture frame, and vice versa, the lower the similarity, the higher the distortion degree, in the case that the distortion degree is greater than the preset degree, a detection report reflecting the drawing quality of the to-be-detected software is generated according to the corresponding detection picture frame and the distortion degree, so that the detection personnel can determine the image drawing quality of the to-be-detected software according to the detection report, and the respective model analysis log is generated according to the respective model running record of the semantic analysis model and the semantic prediction model, so that when the detection personnel needs to compound the detection result, the drawing anomaly caused by model error can be excluded by checking the corresponding model analysis log.

[0086] In the embodiment, by obtaining the distortion degree of each detection picture frame presented when the to-be-detected software draws the to-be-detected video, a detection report reflecting the drawing quality of the to-be-detected software is generated according to the detection picture frame whose distortion degree is greater than the preset degree and the corresponding distortion degree, and the respective model analysis log is generated according to the respective model running record of the semantic analysis model and the semantic prediction model, so that the detection of the drawing quality of the to-be-detected software can be effectively realized, and the normal operation of the to-be-detected software is ensured.

[0087] In an exemplary embodiment, the step of obtaining the current frame detection semantic information of the current detection picture frame includes: identifying the current detection picture frame by the semantic analysis model to obtain recognized text; performing word segmentation on the recognized text, and extracting keywords from the word segmentation result to obtain the current frame detection semantic information.

[0088] Wherein, in order to facilitate the calculation of word vectors and exclude irrelevant information interference, when the current detection picture frame is identified by the semantic analysis model, the recognized text obtained to describe the current detection picture frame can be segmented by the semantic analysis model, and keywords can be extracted from the segmentation result to obtain the current frame detection semantic information, for example, the recognized text is: a person holds a black chopstick from a black bowl to pick up noodles, and the current frame detection semantic information after word segmentation can be represented as: person, hand, black chopstick, black bowl, pick up, noodles.

[0089] In the embodiment, the current detection picture frame is identified by the semantic analysis model to obtain recognized text, the recognized text is segmented, and keywords are extracted from the segmentation result to obtain the current frame detection semantic information, which can simplify the data processing process, exclude irrelevant information interference, and thus improve the detection efficiency of the image drawing quality.

[0090] In an exemplary embodiment, as Figure 5 shown, an image drawing quality detection method is provided, including:

[0091] An action event axis of the to-be-detected video is acquired, and the action event axis is used to indicate at least one action event involved in the to-be-detected video and a corresponding picture frame range of each action event.

[0092] As shown in Figure 6 , the action event axis is represented in the form of a time axis t, and each action event is represented as X1, X2, X3, X4, X5, and the like. Figure 6 As shown in , the action event axis is represented in the form of a time axis t, and each action event is represented as X1, X2, X3, X4, X5, and the like.

[0093] A current picture frame range corresponding to a current action event to which a current detection picture frame of the to-be-detected video belongs is determined.

[0094] As shown in Figure 6 , it is assumed that the current detection picture frame is n1, and the current picture frame range corresponding to the current action event to which n1 belongs can be determined through the action event axis.

[0095] The current detection picture frame is recognized through a semantic analysis model to obtain recognized text, the recognized text is segmented, and keywords are extracted from the segmentation result to obtain current frame detection semantic information; and the detection semantic information of the picture frame after the current detection picture frame in the current picture frame range is predicted according to the current frame detection semantic information.

[0096] The current frame detection semantic information and the predicted detection semantic information are integrated to obtain current detection semantic information; the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame in the current picture frame range are integrated to obtain current standard semantic information; and the similarity between the current detection semantic information and the current standard semantic information is obtained as the similarity between the current detection picture frame and the current standard picture frame; or

[0097] In the current picture frame range, a picture frame comparison range determined by the current detection picture frame and the detection picture frame after the current detection picture frame is acquired; for each pair of detection picture frame and standard picture frame with the same frame sequence in the picture frame comparison range, the similarity between the detection semantic information of the detection picture frame and the standard semantic information of the standard picture frame in each pair of detection picture frame and standard picture frame is acquired; and the similarity between the current detection picture frame and the current standard picture frame is determined according to the similarity corresponding to each pair of detection picture frame and standard picture frame in the picture frame comparison range; and the standard picture frame is presented by the standard software when drawing the to-be-detected video.

[0098] As shown in Figure 6As shown, the picture frame after the current standard picture frame in the current picture frame range is Figure 6 The picture frame corresponding to n1 to t1, and the picture frame comparison range is B: n1~t1.

[0099] After obtaining the standard semantic information of the standard picture frame, the super setting information is obtained; the standard semantic information is verified based on the super setting information, and the standard semantic information that does not pass the verification is modified based on the super setting information.

[0100] According to the similarity between the current detection picture frame and the current standard picture frame, the drawing distortion degree of the to-be-detected software is determined.

[0101] The distortion degree of each detection picture frame presented when the to-be-detected software draws the to-be-detected video is obtained, and the distortion degree of each detection picture frame is determined according to the similarity between each detection picture frame and the corresponding standard picture frame; according to the detection picture frame whose distortion degree is greater than the preset degree and the corresponding distortion degree, a detection report reflecting the drawing quality of the to-be-detected software is generated, and a model analysis log corresponding to each is generated according to the model running record of the semantic analysis model and the semantic prediction model.

[0102] In this embodiment, by obtaining the standard picture frame presented by the standard software drawing the to-be-detected video, for the current detection picture frame presented when the to-be-detected software draws the to-be-detected video, the current frame detection semantic information of the current detection picture frame is obtained, and the detection semantic information of the picture frame after the current detection picture frame is predicted according to the current frame detection semantic information, so as to obtain the similarity between the current detection picture frame and the current standard picture frame according to the standard semantic information corresponding to the standard picture frame and the detection semantic information, and determine the drawing distortion degree of the to-be-detected software. The semantic information in the to-be-detected video can be automatically recognized, and the drawing quality of the to-be-detected software can be detected, so as to improve the detection efficiency of the image drawing quality.

[0103] It should be understood that although each step in the flowchart involved in the above embodiments is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other sequences. Moreover, at least part of the steps in the flowchart involved in the above embodiments can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or steps or stages in other steps.

[0104] Based on the same inventive concept, the embodiments of the present application also provide an image rendering quality detection device for implementing the above-mentioned image rendering quality detection method. The implementation scheme of the device for solving the problem is similar to the implementation scheme described in the above-mentioned method, and therefore the specific limitations in one or more image rendering quality detection device embodiments provided below can refer to the limitations of the image rendering quality detection method described above, which will not be described here again.

[0105] In one exemplary embodiment, as shown in Figure 7 an image rendering quality detection device is provided, comprising a behavior acquisition module 10, a range determination module 20, a semantic prediction module 30, a similarity acquisition module 40 and a distortion determination module 50, wherein:

[0106] The behavior acquisition module 10 is configured to acquire a behavior event axis of a to-be-detected video, the behavior event axis being configured to indicate at least one behavior event involved in the to-be-detected video and a corresponding picture frame range of each behavior event.

[0107] The range determination module 20 is configured to determine, for a current detection picture frame presented by the to-be-detected software when rendering the to-be-detected video, a current picture frame range corresponding to a current behavior event to which the current detection picture frame belongs.

[0108] The semantic prediction module 30 is configured to acquire current frame detection semantic information of the current detection picture frame, and predict detection semantic information of picture frames after the current detection picture frame in the current picture frame range according to the current frame detection semantic information.

[0109] The similarity acquisition module 40 is configured to acquire a similarity between the current detection picture frame and a current standard picture frame according to the current frame detection semantic information and the predicted detection semantic information, and according to current frame standard semantic information of the current standard picture frame and standard semantic information of picture frames after the current standard picture frame in the current picture frame range, the standard picture frame being presented by a standard software when rendering the to-be-detected video.

[0110] The distortion determination module 50 is configured to determine a rendering distortion degree of the to-be-detected software according to the similarity between the current detection picture frame and the current standard picture frame.

[0111] In one exemplary embodiment, the similarity acquisition module 40 is further configured to acquire over-setting information after acquiring the standard semantic information of the standard picture frame; verify the standard semantic information based on the over-setting information, and modify the standard semantic information that fails to pass the verification based on the over-setting information in a case where the verification fails.

[0112] In an exemplary embodiment, the similarity acquisition module 40 is also used to integrate the current frame detection semantic information and the predicted detection semantic information to obtain the current detection semantic information; integrate the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame within the current picture frame range to obtain the current standard semantic information; obtain the similarity between the current detection semantic information and the current standard semantic information as the similarity between the current detection picture frame and the current standard picture frame.

[0113] In an exemplary embodiment, the similarity acquisition module 40 is also used to obtain a picture frame comparison range determined by the current detection picture frame and the detection picture frame after the current detection picture frame within the current picture frame range; for each pair of detection picture frames and standard picture frames with the same frame sequence within the picture frame comparison range, obtain the similarity between the detection semantic information of the detection picture frame and the standard semantic information of the standard picture frame in each pair of detection picture frames and standard picture frames; determine the similarity between the current detection picture frame and the current standard picture frame based on the corresponding similarity of each pair of detection picture frames and standard picture frames within the picture frame comparison range.

[0114] In an exemplary embodiment, the distortion determination module 50 is also used to obtain the degree of distortion of each detection picture frame presented when the software to be detected draws the video to be detected. The degree of distortion of each detection picture frame is determined based on the similarity between each detection picture frame and the corresponding standard picture frame; based on the detection picture frames with a distortion degree greater than a preset degree and the corresponding distortion degree, a detection report reflecting the drawing quality of the software to be detected is generated, and based on the model operation records of the semantic analysis model and the semantic prediction model, corresponding model analysis logs are generated.

[0115] In an exemplary embodiment, the semantic prediction module 30 is also used to identify the current detection picture frame through a semantic analysis model to obtain recognized text; segment the recognized text, and extract keywords from the segmentation results to obtain semantic information of the current frame detection.

[0116] Each module in the above-mentioned image rendering quality detection device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in the computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0117] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 8The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. Among them, the processor, the memory and the input / output interface are connected through a system bus, the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capability. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be realized through WIFI, mobile cellular network, NFC (near field communication) or other technologies. The computer program is executed by the processor to realize an image rendering quality detection method. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0118] Those skilled in the art can understand that, Figure 8 The skilled in the art can understand that,

[0119] In one example embodiment, a computer device is provided, comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program: obtaining an action event axis of a video to be detected, the action event axis being used to indicate at least one action event involved in the video to be detected and a corresponding picture frame range of each action event; determining, for a current detection picture frame presented by a to-be-detected software when rendering the video to be detected, a current picture frame range corresponding to a current action event to which the current detection picture frame belongs; obtaining current frame detection semantic information of the current detection picture frame, and predicting detection semantic information of picture frames after the current detection picture frame in the current picture frame range according to the current frame detection semantic information; obtaining a similarity between the current detection picture frame and a current standard picture frame according to the current frame detection semantic information and the predicted detection semantic information, and current frame standard semantic information of the current standard picture frame and standard semantic information of picture frames after the current standard picture frame in the current picture frame range, the current standard picture frame being presented by a standard software when rendering the video to be detected; and determining a rendering distortion degree of the to-be-detected software according to the similarity between the current detection picture frame and the current standard picture frame.

[0120] In one embodiment, the processor further implements the following steps when executing the computer program: after obtaining the standard semantic information of the standard picture frame, obtaining super setting information; verifying the standard semantic information based on the super setting information, and modifying the standard semantic information that fails to pass the verification based on the super setting information in a case where the verification fails.

[0121] In one embodiment, the processor further implements the following steps when executing the computer program: after obtaining the standard semantic information of the standard picture frame, obtaining super setting information; verifying the standard semantic information based on the super setting information, and modifying the standard semantic information that fails to pass the verification based on the super setting information in a case where the verification fails.

[0122] In one embodiment, the processor, when executing the computer program, involves obtaining similarity between the current detection picture frame and the current standard picture frame according to the detected semantic information obtained from the current frame and the predicted semantic information, and the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame within the current picture frame range, including: obtaining a picture frame comparison range determined by the current detection picture frame and the detection picture frame after the current detection picture frame within the current picture frame range; obtaining similarity between the detection semantic information of the detection picture frame and the standard semantic information of the standard picture frame in each pair of detection picture frame and standard picture frame with consistent frame sequence within the picture frame comparison range; and determining the similarity between the current detection picture frame and the current standard picture frame according to the similarity corresponding to each pair of detection picture frame and standard picture frame within the picture frame comparison range.

[0123] In one embodiment, the process of identifying semantic information is realized by a semantic analysis model, and the process of predicting semantic information is realized by a semantic prediction model; the processor, when executing the computer program, further implements the following steps: obtaining a distortion degree of each detection picture frame presented when the to-be-detected software draws the to-be-detected video, the distortion degree of each detection picture frame being determined according to similarity between each detection picture frame and a corresponding standard picture frame; generating a detection report reflecting drawing quality of the to-be-detected software according to the detection picture frame with a distortion degree greater than a preset degree and the corresponding distortion degree; and generating a model analysis log corresponding to each of the semantic analysis model and the semantic prediction model according to a model running record of each of the semantic analysis model and the semantic prediction model.

[0124] In one embodiment, the processor, when executing the computer program, involves obtaining the current frame detection semantic information of the current detection picture frame, including: identifying the current detection picture frame by the semantic analysis model to obtain identified text; performing word segmentation on the identified text, and extracting keywords from the word segmentation result to obtain the current frame detection semantic information.

[0125] In one embodiment, a computer readable storage medium is provided, and the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the following steps: obtaining a behavior event axis of a video to be detected, the behavior event axis being used to indicate at least one behavior event involved in the video to be detected and a corresponding picture frame range of each behavior event; determining a corresponding current picture frame range of a current behavior event to which a current detection picture frame presented by the video to be detected drawn by a software to be detected belongs; obtaining current frame detection semantic information of the current detection picture frame, and predicting detection semantic information of picture frames after the current detection picture frame in the current picture frame range according to the current frame detection semantic information; obtaining a similarity between the current detection picture frame and a current standard picture frame according to the current frame detection semantic information and the predicted detection semantic information, and current frame standard semantic information of the current standard picture frame and standard semantic information of picture frames after the current standard picture frame in the current picture frame range, the current standard picture frame being presented by the video to be detected drawn by a standard software; and determining a drawing distortion degree of the software to be detected according to the similarity between the current detection picture frame and the current standard picture frame.

[0126] In one embodiment, the computer program is executed by the processor to further implement the following steps: after obtaining the standard semantic information of the standard picture frame, obtaining super setting information; verifying the standard semantic information based on the super setting information, and modifying the standard semantic information that fails to pass the verification based on the super setting information in a case that the standard semantic information fails to pass the verification.

[0127] In one embodiment, the computer program is executed by the processor to further implement the following steps: after obtaining the standard semantic information of the standard picture frame, obtaining super setting information; verifying the standard semantic information based on the super setting information, and modifying the standard semantic information that fails to pass the verification based on the super setting information in a case that the standard semantic information fails to pass the verification.

[0128] In one embodiment, the computer program, when executed by the processor, involves obtaining similarity between the current detection picture frame and the current standard picture frame according to the detected semantic information obtained from the current frame and the predicted semantic information, and the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame within the current picture frame range, including: obtaining a picture frame comparison range determined by the current detection picture frame and the detection picture frame after the current detection picture frame within the current picture frame range; obtaining similarity between the detected semantic information of the detection picture frame and the standard semantic information of the standard picture frame in each pair of detection picture frame and standard picture frame with consistent frame sequence within the picture frame comparison range; and determining the similarity between the current detection picture frame and the current standard picture frame according to the similarity corresponding to each pair of detection picture frame and standard picture frame within the picture frame comparison range.

[0129] In one embodiment, the process of identifying semantic information is realized by a semantic analysis model, and the process of predicting semantic information is realized by a semantic prediction model; the computer program, when executed by the processor, further involves the following steps: obtaining a distortion degree of each detection picture frame presented when the to-be-detected software draws the to-be-detected video, the distortion degree of each detection picture frame being determined according to similarity between each detection picture frame and a corresponding standard picture frame; generating a detection report reflecting drawing quality of the to-be-detected software according to the detection picture frame with a distortion degree greater than a preset degree and the corresponding distortion degree; and generating a model analysis log corresponding to each of the semantic analysis model and the semantic prediction model according to a model running record of each of the semantic analysis model and the semantic prediction model.

[0130] In one embodiment, the computer program, when executed by the processor, involves obtaining the current frame detection semantic information of the current detection picture frame, including: identifying the current detection picture frame by the semantic analysis model to obtain identified text; performing word segmentation on the identified text, and extracting keywords from the word segmentation result to obtain the current frame detection semantic information.

[0131] In one embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the following steps: obtaining an action event axis of a video to be detected, the action event axis being used to indicate at least one action event involved in the video to be detected and a corresponding picture frame range of each action event; determining, for a current detection picture frame presented by a to-be-detected software when drawing the video to be detected, a current picture frame range corresponding to a current action event to which the current detection picture frame belongs; obtaining current frame detection semantic information of the current detection picture frame, and predicting detection semantic information of picture frames after the current detection picture frame in the current picture frame range according to the current frame detection semantic information; obtaining a similarity between the current detection picture frame and a current standard picture frame according to the current frame detection semantic information and the predicted detection semantic information, and current frame standard semantic information of the current standard picture frame and standard semantic information of picture frames after the current standard picture frame in the current picture frame range, the current standard picture frame being presented by a standard software when drawing the video to be detected; and determining a drawing distortion degree of the to-be-detected software according to the similarity between the current detection picture frame and the current standard picture frame.

[0132] In one embodiment, the computer program, when executed by the processor, further implements the following steps: after obtaining the standard semantic information of the standard picture frame, obtaining super setting information; verifying the standard semantic information based on the super setting information, and modifying the standard semantic information that fails to pass the verification based on the super setting information in a case where the verification fails.

[0133] In one embodiment, the computer program, when executed by the processor, involves obtaining the similarity between the current detection picture frame and the current standard picture frame according to the current frame detection semantic information and the predicted detection semantic information, and the current frame standard semantic information of the current standard picture frame and the standard semantic information of picture frames after the current standard picture frame in the current picture frame range, including: integrating the current frame detection semantic information and the predicted detection semantic information to obtain current detection semantic information; integrating the current frame standard semantic information of the current standard picture frame and the standard semantic information of picture frames after the current standard picture frame in the current picture frame range to obtain current standard semantic information; and obtaining a similarity between the current detection semantic information and the current standard semantic information as the similarity between the current detection picture frame and the current standard picture frame.

[0134] In one embodiment, the computer program, when executed by the processor, involves obtaining similarity between the current detection picture frame and the current standard picture frame according to the detected semantic information obtained from the current frame and the predicted semantic information, and the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frame after the current standard picture frame within the current picture frame range, including: obtaining a picture frame comparison range determined by the current detection picture frame and the detection picture frame after the current detection picture frame within the current picture frame range; obtaining similarity between the detected semantic information of the detection picture frame and the standard semantic information of the standard picture frame in each pair of detection picture frame and standard picture frame with consistent frame sequence within the picture frame comparison range; and determining the similarity between the current detection picture frame and the current standard picture frame according to the similarity corresponding to each pair of detection picture frame and standard picture frame within the picture frame comparison range.

[0135] In one embodiment, the process of identifying semantic information is realized by a semantic analysis model, and the process of predicting semantic information is realized by a semantic prediction model; the computer program, when executed by the processor, further involves the following steps: obtaining a distortion degree of each detection picture frame presented when the to-be-detected software draws the to-be-detected video, the distortion degree of each detection picture frame being determined according to similarity between each detection picture frame and a corresponding standard picture frame; generating a detection report reflecting drawing quality of the to-be-detected software according to the detection picture frame with a distortion degree greater than a preset degree and the corresponding distortion degree; and generating a model analysis log corresponding to each of the semantic analysis model and the semantic prediction model according to a model running record of each of the semantic analysis model and the semantic prediction model.

[0136] In one embodiment, the computer program, when executed by the processor, involves obtaining the current frame detection semantic information of the current detection picture frame, including: identifying the current detection picture frame by the semantic analysis model to obtain identified text; performing word segmentation on the identified text, and extracting keywords from the word segmentation result to obtain the current frame detection semantic information.

[0137] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0138] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist, it should be considered as the scope of the present application.

[0139] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A method for detecting image rendering quality, characterized in that: The method comprises: Obtaining a behavior event axis of a video to be detected, where the behavior event axis is used to indicate at least one behavior event involved in the video to be detected and a frame range corresponding to each behavior event; Determine, for a current detection frame presented when the software to be detected draws the video to be detected, a current frame range corresponding to a current behavior event to which the current detection frame belongs; Acquire current frame detection semantic information of the current detection picture frame, and predict detection semantic information of picture frames after the current detection picture frame within the current picture frame range based on the current frame detection semantic information; Obtaining a similarity between the current detection frame and the current standard frame based on the detection semantic information of the current frame and the predicted detection semantic information, as well as the current frame standard semantic information of the current standard frame and the standard semantic information of frames following the current standard frame within the current frame range, where the standard frame is rendered by standard software to present the video to be detected; The rendering distortion degree of the software to be tested is determined according to the similarity between the current detection picture frame and the current standard picture frame.

2. The method according to claim 1, characterized in that The method further comprises: After obtaining standard semantic information of the standard picture frame, obtaining super-setting information; The standard semantic information is verified based on the super-set information, and if the verification fails, the standard semantic information that fails the verification is modified based on the super-set information.

3. The method according to claim 1, characterized in that The obtaining of the similarity between the current detection frame and the current standard frame based on the current frame detection semantic information and the predicted detection semantic information, the current frame standard semantic information of the current standard frame, and the standard semantic information of the frames following the current standard frame within the current frame range includes: Integrate the current frame detection semantic information and the predicted detection semantic information to obtain the current detection semantic information; Integrating the current frame standard semantic information of the current standard picture frame and the standard semantic information of the picture frames after the current standard picture frame within the current picture frame range to obtain the current standard semantic information; The similarity between the current detection semantic information and the current standard semantic information is obtained as the similarity between the current detection picture frame and the current standard picture frame.

4. The method according to claim 1, wherein The obtaining of the similarity between the current detection frame and the current standard frame based on the current frame detection semantic information and the predicted detection semantic information, the current frame standard semantic information of the current standard frame, and the standard semantic information of the frames following the current standard frame within the current frame range includes: Within the current picture frame range, obtaining a picture frame comparison range determined by the current detection picture frame and a detection picture frame subsequent to the current detection picture frame; For each pair of detection picture frames and standard picture frames with the same frame sequence within the picture frame comparison range, obtaining the similarity between the detection semantic information of the detection picture frame and the standard semantic information of the standard picture frame in each pair of detection picture frames and standard picture frames; The similarity between the current detection picture frame and the current standard picture frame is determined according to the similarity corresponding to each pair of detection picture frames and the standard picture frame within the picture frame comparison range.

5. The method according to claim 1, characterized in that The process of identifying semantic information is achieved through a semantic analysis model, and the process of predicting semantic information is achieved through a semantic prediction model; the method further includes: Obtaining a degree of distortion of each detection frame presented when the detection software draws the detection video, wherein the degree of distortion of each detection frame is determined based on a similarity between each detection frame and a corresponding standard frame; Based on the detection picture frames with a distortion degree greater than a preset degree and the corresponding distortion degree, a detection report reflecting the drawing quality of the software to be detected is generated, and based on the respective model operation records of the semantic analysis model and the semantic prediction model, corresponding model analysis logs are generated.

6. The method according to claim 1, characterized in that The obtaining of the current frame detection semantic information of the current detection picture frame includes: Recognize the current detection picture frame through a semantic analysis model to obtain recognized text; The recognized text is segmented, and keywords are extracted from the segmentation results to obtain semantic information of the current frame detection.

7. An image rendering quality detection device, characterized in that: The device comprises: A behavior acquisition module is used to acquire a behavior event axis of a video to be detected, wherein the behavior event axis is used to indicate at least one behavior event involved in the video to be detected and a frame range corresponding to each behavior event; A range determination module is used to determine the current frame range corresponding to the current behavior event to which the current frame is subjected, based on the current frame presented when the software to be detected draws the video to be detected; A semantic prediction module is used to obtain the current frame detection semantic information of the current detection picture frame, and predict the detection semantic information of the picture frame after the current detection picture frame within the current picture frame range based on the current frame detection semantic information; a similarity acquisition module, configured to acquire a similarity between a current detection frame and a current standard frame based on the detection semantic information of the current frame and the predicted detection semantic information, as well as the current frame standard semantic information of the current standard frame and the standard semantic information of frames following the current standard frame within the current frame range, wherein the standard frame is rendered by standard software to present the video to be detected; The distortion determination module is used to determine the drawing distortion degree of the software to be detected based on the similarity between the current detection picture frame and the current standard picture frame.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images