Search

CN-122027822-A - Animation detection method, device, electronic equipment, computer readable storage medium and computer program product

CN122027822ACN 122027822 ACN122027822 ACN 122027822ACN-122027822-A

Abstract

The application provides an animation detection method, an animation detection device, electronic equipment, a computer program product and a computer readable storage medium, wherein the method comprises the steps of responding to a detection request for an animation file, analyzing the animation file to obtain skeleton key points of a virtual object at each moment in a target time period, estimating the gesture of the virtual object based on the skeleton key points of the virtual object at each moment in the target time period to obtain a skeleton gesture sequence of the virtual object in the target time period, determining a moving track of the target key points in the target time period based on the skeleton gesture sequence, and detecting abnormality of the animation file based on the moving track. According to the application, the character animation can be automatically detected, and the detection efficiency is improved.

Inventors

  • MENG RAN
  • YIN YINGYING
  • QIU GUANG
  • HUANG JIAHAO
  • ZHOU FENG

Assignees

  • 腾讯科技（深圳）有限公司

Dates

Publication Date
20260512
Application Date
20241111

Claims (18)

  1. 1. An animation detection method, the method comprising: Responding to a detection request for an animation file, and analyzing the animation file to obtain skeleton key points of a virtual object at each moment in a target time period; Estimating the posture of the virtual object based on skeleton key points of the virtual object at each moment in a target time period to obtain a skeleton posture sequence of the virtual object in the target time period; Determining a moving track of a target key point in the target time period based on the skeleton gesture sequence; and carrying out abnormality detection on the animation file based on the moving track.
  2. 2. The method according to claim 1, wherein the parsing the animation file to obtain skeletal keypoints of the virtual object at each moment in the target time period comprises: If the animation file is a data file for storing the position information of the skeleton key point of the virtual object in the target time, extracting an animation frame sequence of the virtual object in the target time period from the data file; determining skeleton key points of the virtual object at each moment in the target time period based on the animation frame sequence; Wherein each animation frame in the animation frame sequence comprises position information of skeleton key points of the virtual object, and one animation frame corresponds to one moment.
  3. 3. The method according to claim 1, wherein the parsing the animation file to obtain skeletal keypoints of the virtual object at each moment in the target time period comprises: If the animation file is a video file interacted by the virtual object in the target time, extracting animation frames from the video file to obtain an animation frame sequence comprising the virtual object, wherein one animation frame corresponds to one moment; And extracting skeleton key points of the virtual objects in the animation frames aiming at each animation frame in the animation frame sequence to obtain the skeleton key points of the virtual objects in each animation frame.
  4. 4. A method according to claim 3, wherein said extracting skeleton key points of the virtual object in the animation frame to obtain skeleton key points of the virtual object in each animation frame comprises: extracting image features of the virtual objects in the animation frames to obtain image features; Performing key point probability prediction on pixel points in the animation frame based on the image characteristics to obtain probability confidence that each pixel point in the animation frame is a skeleton key point; and screening pixel points in the animation frame based on the probability confidence coefficient to obtain skeleton key points of the virtual object in the animation frame.
  5. 5. A method according to claim 2 or 3, wherein estimating the pose of the virtual object based on the skeletal keypoints of the virtual object at each moment in the target time period to obtain the skeletal pose sequence of the virtual object in the target time period comprises: for each animation frame in the target time period, the connection relation between skeleton key points in the animation frame is presumed, and a presumption result is obtained; Establishing a connection relation between the skeleton key points based on the estimation result to obtain skeleton postures of the virtual objects in the animation frames; And arranging the skeleton postures of the virtual objects in the animation frames according to the time sequence of the animation frames to obtain a skeleton posture sequence of the virtual objects in the target time period.
  6. 6. The method of claim 1, wherein the determining a movement trajectory of a target keypoint over the target time period based on the sequence of skeletal poses comprises: Extracting, for each bone pose in the sequence of bone poses, a position of a target keypoint of the bone keypoints in the bone pose; and connecting the positions of the target key points based on the time sequence of the skeleton gesture in the skeleton gesture sequence to obtain the moving track of the target key points in the target time period.
  7. 7. The method according to claim 1, wherein the abnormality detection of the animation file based on the movement trajectory includes: determining a first acceleration of the target key point at a first moment, a second acceleration of the target key point at a second moment, a third acceleration of the target key point at a third moment and a fourth acceleration of the target key point at a fourth moment based on the position of the target key point in the moving track, wherein the first moment, the second moment, the third moment and the fourth moment are four continuous moments in the target time period; Determining an acceleration change rate of the target key point from the first moment to the second moment based on the first acceleration and the second acceleration; Determining that the target key point has movement jam under the condition that the acceleration change rate exceeds a change rate threshold value, the second acceleration exceeds a first acceleration threshold value, and the third acceleration and the fourth acceleration are lower than the first acceleration threshold value; And when the target key point has a mobile click, determining that the virtual object in the animation file has an action click abnormality.
  8. 8. The method according to claim 1, wherein the abnormality detection of the animation file based on the movement trajectory includes: determining the speed of the target key point in at least four moments in the target time period based on the moving track, wherein the at least four moments are discontinuous four moments; determining a speed range of the target key point in the target time period based on the speeds of the target key point at the at least four moments; determining the speed of the target key point at the target moment based on the moving track aiming at the target moment in the target time period; when the speed of the target key point at the target moment exceeds the speed range, determining that the target key point has motion abnormality at the target moment; And when the number of the target key points with motion abnormality in the target time exceeds a number threshold, determining that the animation frames corresponding to the target time in the animation file are abnormal.
  9. 9. The method of claim 1, wherein the step of determining the position of the substrate comprises, The target key points comprise a first key point, a second key point and a third key point, wherein the first key point and the second key point are used for representing the whole virtual object; The detecting the abnormality of the animation file based on the moving track comprises the following steps: determining a first speed, a second speed and a third speed of the first key point, the second key point and the third key point at a target moment respectively based on the movement track; And when the third speed is higher than a first speed threshold value and the first speed and the second speed are both lower than the first speed threshold value, or the third speed does not exceed the first speed threshold value and the first speed is higher than the first speed threshold value or the second speed is higher than the first speed threshold value, determining that the virtual object in the animation file has a sliding abnormality at the target moment.
  10. 10. The method according to claim 1, wherein the abnormality detection of the animation file based on the movement trajectory includes: Determining the vertical position of the target key point in the vertical direction at each moment based on the movement track; and determining an abnormal time when the vertical position of the target key point is lower than a preset position, and determining an animation frame corresponding to the abnormal time in the animation file as an abnormal frame.
  11. 11. The method of claim 1, wherein the target keypoints comprise a third keypoint for characterizing the virtual object as a whole and a fourth keypoint of a designated joint of the virtual object; The detecting the abnormality of the animation file based on the moving track comprises the following steps: Determining a first relative speed of the fourth key point and the third key point at a fifth moment and a second relative speed of at least two sixth moments after the fifth moment based on the movement track; Determining relative accelerations of the fourth and third keypoints at the sixth moments based on the first relative speeds and the second relative speeds; Determining that movement jamming exists between the fifth moment and the sixth moment of the designated joint under the condition that the first relative speed is greater than a second speed threshold, each second relative speed is lower than a third speed threshold and each relative acceleration is lower than a second acceleration threshold; And when the specified joint has a moving click, determining that the virtual object in the animation file has an action click abnormality.
  12. 12. The method according to claim 1, wherein the abnormality detection of the animation file based on the movement trajectory includes: Determining a relative rotation angle of the target key point between a seventh time and an eighth time after the seventh time based on the movement track; and when the relative rotation angle is higher than a rotation angle threshold value, determining that the virtual object in the animation file has rotation abnormality at the seventh moment.
  13. 13. The method according to any one of claims 1 to 12, wherein after the anomaly detection of the animation file, the method further comprises: Displaying at least one of an abnormal detail list, a time sequence diagram, a model view corresponding to the virtual object and a video playing interface of the animation file, wherein the information corresponds to the animation file; the abnormal detail list is used for displaying abnormal information existing in the animation file; The time sequence diagram comprises at least one of a speed time sequence diagram for showing the speed change condition of a target key point of the virtual object, an acceleration time sequence diagram for showing the acceleration change of the target key point, a position time sequence diagram for showing the position change condition of the target key point and an angle time sequence diagram for showing the steering angle change condition of the target key point; the model view is used for displaying the skeleton gesture of the virtual object at the target moment; the video playing interface is used for playing the animation video of the animation file.
  14. 14. The method of claim 13, wherein the time series plot, the video playback interface, and the model view have a time series relationship, the method further comprising: And responding to the dragging operation of the progress bar in the video playing interface, displaying an animation frame corresponding to a target moment in the video playing interface, displaying a skeleton gesture corresponding to the target moment in the model view, and highlighting the speed, acceleration and steering angle corresponding to the target moment in the time sequence diagram.
  15. 15. An animation detection device, the device comprising: The analysis module is used for responding to the detection request for the animation file, analyzing the animation file and obtaining skeleton key points of the virtual object at each moment in the target time period; The gesture estimation module is used for estimating the gesture of the virtual object based on the skeleton key points of the virtual object at each moment in a target time period to obtain a skeleton gesture sequence of the virtual object in the target time period; The determining module is used for determining the moving track of the target key point in the target time period based on the skeleton gesture sequence; and the detection module is used for detecting the abnormality of the animation file based on the movement track.
  16. 16. An electronic device, the electronic device comprising: A memory for storing computer executable instructions or computer programs; a processor for implementing the method of any one of claims 1 to 14 when executing computer-executable instructions or computer programs stored in the memory.
  17. 17. A computer-readable storage medium storing computer-executable instructions or a computer program, which when executed by a processor implement the method of any one of claims 1 to 14.
  18. 18. A computer program product comprising computer-executable instructions or a computer program, which, when executed by a processor, implements the method of any one of claims 1 to 14.

Description

Animation detection method, device, electronic equipment, computer readable storage medium and computer program product Technical Field The present application relates to animation technology, and in particular, to an animation detection method, an animation detection device, an electronic device, a computer readable storage medium, and a computer program product. Background Animation detection plays a very important role in the animation production process, and problems in animation production can be timely found and solved through detection, so that the quality of animation works is ensured, and the user experience is optimized. For animation detection, there are detection modes of automatically generating test cases and detection modes of pre-configuring test files in the related art, however, these modes in the related art cannot be suitable for detecting the character animation, and if the character animation is detected by the traditional manual mode, time and effort are consumed, and efficiency is low. Disclosure of Invention The embodiment of the application provides an animation detection method, an animation detection device, electronic equipment, a computer readable storage medium and a computer program product, which can automatically detect character animations and improve detection efficiency. The technical scheme of the embodiment of the application is realized as follows: The embodiment of the application provides an animation detection method, which comprises the following steps: The method comprises the steps of responding to a detection request aiming at an animation file, analyzing the animation file to obtain skeleton key points of a virtual object at each moment in a target time period, estimating the gesture of the virtual object based on the skeleton key points of the virtual object at each moment in the target time period to obtain a skeleton gesture sequence of the virtual object in the target time period, determining a moving track of the target key points in the target time period based on the skeleton gesture sequence, and detecting abnormality of the animation file based on the moving track. The embodiment of the application provides an animation detection device, which comprises: The analysis module is used for responding to the detection request for the animation file, analyzing the animation file and obtaining skeleton key points of the virtual object at each moment in the target time period; The gesture estimation module is used for estimating the gesture of the virtual object based on the skeleton key points of the virtual object at each moment in a target time period to obtain a skeleton gesture sequence of the virtual object in the target time period; The determining module is used for determining the moving track of the target key point in the target time period based on the skeleton gesture sequence; and the detection module is used for detecting the abnormality of the animation file based on the movement track. In the above scheme, the analyzing module is further configured to extract an animation frame sequence of the virtual object in the target time period from the data file if the animation file is a data file for storing the position information of the skeletal key point of the virtual object in the target time period, determine the skeletal key point of the virtual object at each moment in the target time period based on the animation frame sequence, wherein each animation frame in the animation frame sequence includes the position information of the skeletal key point of the virtual object, and one animation frame corresponds to one moment. In the above scheme, the analysis module is further configured to extract an animation frame from the video file if the animation file is a video file in which the virtual object interacts within a target time, to obtain an animation frame sequence including the virtual object, wherein one animation frame corresponds to one time, and extract a skeleton key point from the virtual object in the animation frame for each animation frame in the animation frame sequence, to obtain the skeleton key point of the virtual object in each animation frame. In the scheme, the analysis module is further used for extracting image features of the virtual object in the animation frame to obtain image features, conducting key point probability prediction on the pixel points in the animation frame based on the image features to obtain probability confidence that each pixel point in the animation frame is a skeleton key point, and screening the pixel points in the animation frame based on the probability confidence to obtain the skeleton key point of the virtual object in the animation frame. In the above scheme, the gesture estimation module is further configured to infer a connection relationship between skeleton key points in the animation frames for each animation frame in the target time period to obtain an inference result, establish a con