Challenge Introduction

The lack of large-scale data severely hinders the progress of sign language recognition (SLR), as manually collecting and labeling sign videos is rather time-consuming. To facilitate SLR, we release a new large-scale Chinese dataset, which can generate a large number of samples in a novel cost- controlled manner. With only a few sub-videos, hundreds of videos could be synthesized by combining different sub-video pairs. Inference is performed on normally recorded videos.
This competition hopes to demonstrate the effectiveness of large-scale data on SLR, i.e., training on the combinatorial data and testing on the complete data.

Challenge Organization

Challenge Co-Chairs

Organizing Team

  • Wei Feng, Tianjin University, China
  • Di Lin, Tianjin University, China
  • Qing Guo, Agency for Science, Technology and Research (A*STAR), Singapore
  • Zhensong Zhang, Huawei, China

Challenge Webpage


Challenge Contact

Di Lin, Tianjin University, China