Rethinking Token Reduction with Parameter-Efficient Fine-Tuning in ViT for Pixel-Level Tasks.

Cheng Lei, Ao Li, Hu Yao, Ce Zhu, Le Zhang

Browse the full CVPR paper archive.