Research map / Video / Video segmentation and tracking
Spatio-Temporal Gating-Adjacency GCN for Human Motion Prediction
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use GAGCN instead of STSGCN | Human 3.6M · Human 3.6M average, 1000ms | MPJPE: 77.3 -> 72.9 | Helped |
| Use GAGCN instead of STSGCN | AMASS-BMLrub · AMASS-BMLrub average, 1000ms | MPJPE: 45.5 -> 43.1 | Helped |
| Use GAGCN instead of STSGCN | 3DPW · 3DPW average, 1000ms | MPJPE: 42.3 -> 39.9 | Helped |
| Use GAGCN instead of STSGCN | Human 3.6M angle-based prediction · Human 3.6M average, 1000ms | MAE: 1.09 -> 1.02 | Helped |
| Replace stable spatial and temporal adjacency matrices with GAGCN's adaptive gating (ablation) | Human 3.6M Walking Together unseen · Human 3.6M Walking Together unseen, 1000ms | MPJPE: 69.8 -> 51.6 | Helped |
| Train with Walking Together seen rather than unseen (ablation) | GAGCN · Human 3.6M Walking Together, 1000ms | MPJPE: 51.6 -> 50.4 | No clear effect |
| Add gating over temporal candidate adjacency matrices (ablation) | GAGCN · Human 3.6M average, 1000ms | MPJPE: 93.9 -> 72.9 | Helped |
| Add gating over spatial candidate adjacency matrices (ablation) | GAGCN · Human 3.6M average, 1000ms | MPJPE: 91.1 -> 72.9 | Helped |
| Increase candidate adjacency counts from S4, T3 to S8, T6 (ablation) | GAGCN · Human 3.6M average, 1000ms | MPJPE: 72.9 -> 76.9 | Hurt |
| Change candidate adjacency counts from S4, T3 to S3, T4 (ablation) | GAGCN · Human 3.6M average, 1000ms | MPJPE: 72.9 -> 75.3 | Hurt |
Closest papers by meaning
- GCNext: Towards the Unity of Graph Convolutions for Human Motion Prediction (AAAI 2024 · similarity 0.89)
- Spatio-Temporal Fusion for Human Action Recognition via Joint Trajectory Graph (AAAI 2024 · similarity 0.87)
- Structural-RNN: Deep Learning on Spatio-Temporal Graphs (CVPR 2016 · similarity 0.87)
- Are Spatial-Temporal Graph Convolution Networks for Human Action Recognition Over-Parameterized? (CVPR 2025 · similarity 0.87)
- Progressively Generating Better Initial Guesses Towards Next Stages for High-Quality Human Motion Prediction (CVPR 2022 · similarity 0.86)
- BlockGCN: Redefine Topology Awareness for Skeleton-Based Action Recognition (CVPR 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.