3D Hand Poses for Efficient Action Recognition: HandFormer Approach

Efficiently recognizing actions using 3D hand poses with the HandFormer multimodal transformer.