计算机与数字工程2025，Vol.53Issue(1)：263-268,6.DOI:10.3969/j.issn.1672-9722.2025.01.047
基于姿态识别的马体视频关键帧提取方法
Keyframe Extraction Method of Horse Body Video Based on Posture Recognition
摘要
Abstract
This study focuses on horse contour segmentation,pose information acquisition,and pose classification methods to provide a feasible solution for finding video key frames suitable for key measurement information.The BlendMask network is select-ed for instance segmentation of video frame,and then the Canny operator is used to extract the horse contour information from the segmented image,and then the horse posture information is obtained.Finally,the SVM classifier is used to classify the posture to obtain the key frame suitable for measuring the horse key information.The experimental results show that the key frame selection method suitable for horse key information measurement based on contour information can obtain the standing key frame suitable for horse key information measurement with an accuracy of nearly 97%.关键词
动物姿态识别/BlendMask/轮廓信息获取/SVM/CannyKey words
animal posture recognition/BlendMask/contour information acquisition/SVM/Canny分类
信息技术与安全科学引用本文复制引用
贾东霖,张婧婧,李全胜,周腾飞,王迎超..基于姿态识别的马体视频关键帧提取方法[J].计算机与数字工程,2025,53(1):263-268,6.基金项目
国家自然科学基金项目(编号:32060321) （编号:32060321）
新疆维吾尔自治区高校科研计划项目(编号:XJEDU2020Y020) （编号:XJEDU2020Y020）
新疆农业大学2022年度自治区级研究生科研创新项目(编号:XJ2022G138) （编号:XJ2022G138）
新疆农业大学研究生教育教学改革研究项目(编号:xjaualk-yjs-2021012) （编号:xjaualk-yjs-2021012）
新疆农业大学2022年度研究生科研创新项目(编号:XJAUGRI2022033) （编号:XJAUGRI2022033）
三亚学院中青年教师专项培养项目(科研类)(编号:USYJSPY24-51)资助. （科研类）
