A method for outdoor group emotion recognition based on skeleton keypoints is proposed.Firstly,the YOLOPose algorithm is employed for tracking,detecting,and pose estimation on each individual in the group.Secondly,the skeleton information is obtained through pose estimation,and the emotion features of each person are computed based on the skeleton information.Finally,the skeleton information and emotion features are input into the established dual-branch pose-emotion recognition model(D-ConvLSTM),extracts and fuses multi-level spatiotemporal features from both branches,so as to perform emotion classification according to fused dual-branch features,obtain the emotion recognition results for each person in the group.Experimental results show that the method has high recognition precision in outdoor group emotion recognition.
关键词
室外群体情绪识别/骨骼关键点/特征融合/D-ConvLSTM模型
Key words
outdoor group emotion recognition/skeleton keypoints/feature fusion/D-ConvLSTM model
维普
万方数据