Share your thoughts, 1 month free Claude Pro on usSee more

Emotion Recognition on MER-UniBench (test)

78.54MER23

AffectGPT

Updated 1mo ago

Evaluation Results

Method	Links
AffectGPT 2026.01		78.54	78.8	55.65	60.54	81.3	80.9	88.49	86.18	62.52	74.77
Emotion-LLaMAv2 2026.01		77.28	86.9	51.32	84.05	86.28	87.69	87.72	86.02	62.9	78.91
Emotion-LLaMAv2 2026.01		76.72	86.9	48.61	83.6	86.96	86.15	86.35	87.42	64	78.52
Qwen2.5-Omni-7B 2026.01		72.84	80.44	43.72	61.53	78.86	61.84	75.44	76.67	53.13	67.16
HumanOmniV2-9B 2026.01		68.59	57.41	42.74	60.95	76.56	75.44	83.58	81.65	61.62	67.62
HumanOmni-7B 2026.01		63.36	76.64	37.05	52.52	7.37	7.33	43.87	35.35	59.03	42.5
Emotion-LLaMA 2026.01		59.38	73.62	46.76	55.47	66.13	67.66	78.32	77.23	52.97	64.17
LLaMA3.2-90B-Vision 2026.01		58.49	62.72	48.82	57.71	82.06	65.74	77.65	80.26	54.09	65.28
Video-LLaVA 2026.01		54.82	43.31	35.96	37.17	71.31	40.81	61.2	57.17	52.5	50.47
Qwen2.5-VL-72B 2026.01		47.84	40.43	46.81	34.86	85.81	70.8	72	72.85	47.02	57.6
Video-LLaMA2 2026.01		47.56	54.59	19.58	59.41	58.06	44.08	63.15	70.63	52.32	52.15
Video-LLaMA 2026.01		38.65	44.4	31.6	20.36	69.19	36.75	31.16	31.21	49.91	39.25