1 paper · 1 filter
Yingshan Chang, Yasi Zhang, Zhiyuan Fang +3
The literature on text-to-image generation is plagued by issues of faithfully composing entities with relations. But there lacks a formal understanding of how entity-relation compo…