Dynamic Token Normalization improves Vision Transformers.

Wenqi Shao Yixiao Ge Zhaoyang Zhang Xuyuan Xu Xiaogang Wang Ying Shan Ping Luo

Published in: ICLR (2022)

Keyphrases

computer vision
real time
dynamic environments
image processing
preprocessing
vision system
management system
databases
learning algorithm
high level
feature extraction
lower bound
special case
normalization method