Rethinking and Improving Relative Position Encoding for Vision Transformer.

Kan Wu, Houwen Peng, Minghao Chen, Jianlong Fu, Hongyang Chao

Browse the full ICCV paper archive.