Research on video activity understanding based on deep learning
Hongjun Chen, Junlian Xiang, Fuqiang Luo, Mingkun Kao, Ruilin Li · 2023
There are very rich individual and group activity in the video. It is very significant to excavate the characteristics of individuals and groups and analyze the interaction between them. In this paper convolution neural network and graph relationship are used to identify the activities of individuals and individuals in the video, and the accuracy rate and intersection ratio are used to evaluate, and good results are obtained after verification.