Exploring Cross-Attention Maps in Multi-modal Diffusion Transformers for Training-Free Semantic Segmentation.

Rento Yamaguchi, Keiji Yanai

Browse the full ACCV paper archive.