notesum.ai

Published at November 18

MEMO-Bench: A Multiple Benchmark for Text-to-Image and Multimodal Large Language Models on Human Emotion Analysis

cs.CL

cs.AI

Released Date: November 18, 2024

Authors: Yingjie Zhou¹, Zicheng Zhang¹, Jiezhang Cao², Jun Jia¹, Yanwei Jiang¹, Farong Wen¹, Xiaohong Liu¹, Xiongkuo Min¹, Guangtao Zhai¹

Aff.: ¹Shanghai Jiaotong University, PengCheng Laboratory; ²Harvard Medical School

Arxiv: http://arxiv.org/abs/2411.11235v1

[Uncaptioned image]

T2I	$GACC$
T2I	HAP	SAD	WOR	NEU	SUR	ANG	$All$
APA	1.0000	0.5900	0.3600	0.6600	0.8300	0.7400	0.6967
DL3	0.9683	0.5500	0.3807	0.7005	0.6667	0.7120	0.7160
FCS	0.9000	0.1200	0.0700	0.8000	0.2800	0.1546	0.3886
FLU	0.9900	0.3300	0.3400	0.6700	0.6000	0.8100	0.6233
KDS	0.9600	0.1600	0.1300	0.6100	0.5200	0.2600	0.4400
ODE	0.9600	0.4600	0.3100	0.6800	0.5600	0.6768	0.6077
PTS	0.9300	0.3500	0.2600	0.8000	0.5400	0.3838	0.5442
SD1	0.9596	0.5556	0.2400	0.6162	0.1919	0.3368	0.4839
SD2	0.9500	0.3838	0.3200	0.5600	0.3918	0.1224	0.4562
SDX	0.9800	0.3700	0.2200	0.7000	0.4200	0.2959	0.4983
SD3	0.9800	0.2300	0.1200	0.6600	0.3300	0.3000	0.4367
SGA	0.9000	0.6000	0.4200	0.7600	0.8700	0.9200	0.7450