Spatio-Temporal Gating-Adjacency GCN for Human Motion Prediction
This work improves human motion prediction for applications in autonomous driving and robotics, but it is incremental as it builds on existing Graph Convolutional Networks with novel gating mechanisms.
The paper tackles the problem of predicting future human motion from historical sequences by addressing the cross-dependency of spatio-temporal relationships, which is challenging due to decoupled modeling and insufficient generalization. The proposed Spatio-Temporal Gating-Adjacency GCN achieves state-of-the-art performance on datasets like Human 3.6M, AMASS, and 3DPW for both short-term and long-term predictions.
Predicting future motion based on historical motion sequence is a fundamental problem in computer vision, and it has wide applications in autonomous driving and robotics. Some recent works have shown that Graph Convolutional Networks(GCN) are instrumental in modeling the relationship between different joints. However, considering the variants and diverse action types in human motion data, the cross-dependency of the spatio-temporal relationships will be difficult to depict due to the decoupled modeling strategy, which may also exacerbate the problem of insufficient generalization. Therefore, we propose the Spatio-Temporal Gating-Adjacency GCN(GAGCN) to learn the complex spatio-temporal dependencies over diverse action types. Specifically, we adopt gating networks to enhance the generalization of GCN via the trainable adaptive adjacency matrix obtained by blending the candidate spatio-temporal adjacency matrices. Moreover, GAGCN addresses the cross-dependency of space and time by balancing the weights of spatio-temporal modeling and fusing the decoupled spatio-temporal features. Extensive experiments on Human 3.6M, AMASS, and 3DPW demonstrate that GAGCN achieves state-of-the-art performance in both short-term and long-term predictions.