A Deep Multimodal Model for Predicting Affective Responses Evoked by Movies Based on Shot Segmentation

Table 4

Comparison of state-of-the-art results for experienced emotion prediction.

Features	Arousal (loss1)		Valence (loss2)
	MSE	PCC	MSE	PCC
All features	0.0275	0.6187	0.0632	0.3443
−Action features	0.0291	0.6038	0.0673	0.3259
−Face features	0.0277	0.6136	0.0637	0.3667
−Person features	0.0280	0.6181	0.0653	0.3726
−Place features	0.0280	0.5981	0.0663	0.3315
−VGGish features	0.0290	0.5952	0.0669	0.3444
−OpenSMILE features	0.0295	0.6003	0.0666	0.3345
All_visual_features	0.0316	0.4931	0.0751	0.2694
All_audio_features	0.0297	0.6141	0.0726	0.3356

“−” indicates without the feature.