Improved TokenPose with Sparsity

Anning Li

Over the past few years, the vision transformer and its various forms have gained significance in human pose estimation. By treating image patches as tokens, transformers can captu…