Hierarchical Attention-Based Multimodal Fusion Network for Video Emotion Recognition

Table 3

Accuracy of emotion recognition of different modals.

Convolution layers	Face features accuracy (%)	Scene features accuracy (%)	Image features accuracy (%)
L1	55.14	44.39	46.03
L2	57.94	44.62	42.99
L3	54.67	43.92	44.62