AI RESEARCH
Rethinking Graph Convolution for 2D-to-3D Hand Pose Lifting
arXiv CS.CV
•
ArXi:2605.13604v1 Announce Type: new Graph convolutional networks (GCNs) are widely used for 3D hand pose estimation, where the hand skeleton is encoded as a fixed adjacency graph. We revisit whether this is the most effective way to incorporate hand topology in 2D-to-3D lifting. In this paper, we perform controlled, parameter-matched ablations on the FPHA benchmark and show that standard multi-head self-attention consistently outperforms GCN baselines.