Research map / Video / Video segmentation and tracking
Multi-Stream Representation Learning for Pedestrian Trajectory Prediction
Yuxuan Wu, Le Wang, Sanping Zhou, Jinghai Duan, Gang Hua, Wei Tang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 three separate temporal, spatial, and cross branches instead of joint flattened self-attention (ablation) | MSRL · ETH-UCY AVG | ADE: 0.26 -> 0.21 | Improved |
| Use three separate temporal, spatial, and cross branches instead of joint flattened self-attention (ablation) | MSRL · ETH-UCY AVG | FDE: 0.43 -> 0.34 | Improved |
| Add a spatial branch to the temporal branch (ablation) | MSRL · ETH-UCY AVG | ADE: 0.24 -> 0.22 | Improved |
| Add a cross-correlation branch to temporal and spatial branches (ablation) | MSRL · ETH-UCY AVG | ADE: 0.22 -> 0.21 | Improved |
| Add gated feature fusion (ablation) | MSRL · ETH-UCY AVG | ADE: 0.21 -> 0.20 | Improved |
| Add sparse attention gates to spatial and cross branches (ablation) | MSRL · ETH-UCY AVG | ADE: 0.20 -> 0.19 | Improved |
| Replace the multi-stream module with a temporal transformer followed by a spatial transformer (ablation) | MSRL · ETH-UCY AVG | ADE: 0.21 -> 0.22 | Hurt |
| Combine spatial and cross interactions in one branch rather than separating them (ablation) | MSRL · ETH-UCY AVG | ADE: 0.21 -> 0.21 | No clear effect |
| Use MSRL rather than LB-EBM | ETH-UCY trajectory prediction · ETH-UCY AVG | ADE: 0.21 -> 0.19 | Improved |
| Use MSRL rather than MemoNet | SDD trajectory prediction · SDD | ADE: 8.56 -> 8.22 | Improved |
| Use MSRL rather than MemoNet | SDD trajectory prediction · SDD | FDE: 12.66 -> 13.39 | Hurt |
Closest papers by meaning
- Complementary Attention Gated Network for Pedestrian Trajectory Prediction (AAAI 2022 · similarity 0.89)
- A Unified Environmental Network for Pedestrian Trajectory Prediction (AAAI 2024 · similarity 0.89)
- STGAT: Modeling Spatial-Temporal Interactions for Human Trajectory Prediction (ICCV 2019 · similarity 0.88)
- TrajCLIP: Pedestrian trajectory prediction method using contrastive learning and idempotent networks (NeurIPS 2024 · similarity 0.86)
- MS-TIP: Imputation Aware Pedestrian Trajectory Prediction (ICML 2024 · similarity 0.86)
- Towards Predicting Any Human Trajectory In Context (NeurIPS 2025 · similarity 0.86)
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.
