notesum.ai

Published at November 27

cs.CV

Released Date: November 27, 2024

Authors: Wataru Shimoda¹, Naoto Inoue¹, Daichi Haraguchi¹, Hayato Mitani², Seichi Uchida², Kota Yamaguchi¹

Aff.: ¹CyberAgent; ²Kyushu University

Method	GPT $\uparrow$		OCR $\uparrow$	FID $\downarrow$	CLIP $\uparrow$
Method	Graphic	Match	OCR $\uparrow$	FID $\downarrow$	CLIP $\uparrow$
TD [9]	4.67	7.16	55.4	42.0	34.4
TD-2 [10]	4.97	7.17	56.2	33.8	34.7
Type-R
SD3[11]	7.30	8.46	48.6	45.0	33.0
Flux[3]	7.67	8.55	62.0	43.1	33.1