Abstract: In 3D skeleton-based action recognition, the limited availability of supervised data has driven interest in self-supervised learning methods. The reconstruction paradigm using masked ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback