Region Dual Attention-Based Video Emotion Recognition

Table 5

Top-1 accuracy (%) compared with related works on the MHED and HEIV.

Method	Accuracy on MEHD (%)	Accuracy on HEIV (%)
Quality-aware network [6]	46.03	43.95
Vielzeuf et al. [13]	53.73	45.93
Chen et al. [12]	55.60	46.17
Attention clusters [7]	59.81	49.63
Our method	65.89	53.09