Jun, 2024
COT Flow:通过对比对学习最优转运图像采样和编辑
COT Flow: Learning Optimal-Transport Image Sampling and Editing by Contrastive Pairs
Xinrui Zu, Qian Tao
TL;DRDiffusion models suffer from computationally expensive iterative generation, limited editing flexibility, and Gaussian noise constraints; Contrastive Optimal Transport Flow (COT Flow) overcomes these limitations, enabling fast, high-quality generation and improved zero-shot editing with unpaired image-to-image translation and doubled editable space.