Efficient Vision Transformers via Token Merging with Head-Wise Attention Correction.

Yuki Ichikawa, Masato Motomura, Thiem Van Chu, Daichi Fujiki

Browse the full WACV paper archive.