This paper is aimed at addressing the influence factors including poor stability and weak generalization ability on human body action video recognition,and proposes a TSM2.0 optimization algo-rithm.This is achieved by using Resnet50 network as the backbone network for feature extraction;intro-ducing non-local operators to optimize TSM network,improve the overall performance of the network mod-el,and optimize the network structure.The results show that the downward state of the loss function curve obtained by TSM2.0 algorithm is more stable,and the training accuracy rate is more than 92%,which effectively reduces the overfitting phenomenon of the loss function curve in the early stage,and im-proves the stability and generalization ability of the model.This research has better performance in human motion video recognition.
关键词
视频识别/深度学习/TSM/非局部操作算子
Key words
video recognition/deep learning/TSM/non-local operator
维普
万方数据