STAR-Transformer: A Spatio-temporal Cross Attention Transformer for Human Action Recognition

Benchmark Model Rank Results
action-recognition-in-videos-on-ntu-rgbd-120STAR-Transformer (RGB + Pose)Accuracy (Cross-Setup): 92.7Accuracy (Cross-Subject): 90.3
action-recognition-in-videos-on-ntu-rgbd-60STAR-Transformer (RGB + Pose)Accuracy (CS): 92.0Accuracy (CV): 96.5