Research map / Architectures and efficiency / RNNs and normalization
Social LSTM: Human Trajectory Prediction in Crowded Spaces
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 |
|---|---|---|---|
| Add social pooling of neighboring hidden states (ablation) | LSTM trajectory predictor · UCY | average displacement error: 0.35 -> 0.27 | Helped |
| Add social pooling of neighboring hidden states (ablation) | LSTM trajectory predictor · UCY | average non-linear displacement error: 0.20 -> 0.16 | Helped |
| Add social pooling of neighboring hidden states (ablation) | LSTM trajectory predictor · UCY | final displacement error: 0.90 -> 0.77 | Helped |
| Pool neighbor coordinates rather than hidden states (ablation) | Social-LSTM · ETH | average displacement error: 0.49 -> 0.50 | No clear effect |
| Pool neighbor coordinates rather than hidden states (ablation) | Social-LSTM · UCY | average displacement error: 0.27 -> 0.35 | Hurt |
| Pool neighbor coordinates rather than hidden states (ablation) | Social-LSTM · UCY | average non-linear displacement error: 0.16 -> 0.20 | Hurt |
| Use social interaction modeling instead of independent per-person LSTMs (ablation) | LSTM trajectory predictor · Average across five datasets | average displacement error: 0.44 -> 0.27 | Helped |
| Use social interaction modeling instead of independent per-person LSTMs (ablation) | LSTM trajectory predictor · Average across five datasets | average non-linear displacement error: 0.24 -> 0.15 | Helped |
| Use social interaction modeling instead of independent per-person LSTMs (ablation) | LSTM trajectory predictor · Average across five datasets | final displacement error: 0.98 -> 0.61 | Helped |
| Use Social-LSTM rather than the strongest listed baseline | trajectory prediction · ETH | average displacement error: 0.20 -> 0.50 | Hurt |
| Use Social-LSTM rather than the strongest listed baseline | trajectory prediction · HOTEL | average displacement error: 0.15 -> 0.11 | Helped |
| Use Social-LSTM rather than the strongest listed baseline | trajectory prediction · ZARA 2 | average displacement error: 0.28 -> 0.25 | Helped |
Closest papers by meaning
- Social-Transmotion: Promptable Human Trajectory Prediction (ICLR 2024 · similarity 0.89)
- SocialCVAE: Predicting Pedestrian Trajectory via Interaction Conditioned Latents (AAAI 2024 · similarity 0.88)
- SocialMOIF: Multi-Order Intention Fusion for Pedestrian Trajectory Prediction (CVPR 2025 · similarity 0.87)
- Following the Human Thread in Social Navigation (ICLR 2025 · similarity 0.86)
- GigaTraj: Predicting Long-term Trajectories of Hundreds of Pedestrians in Gigapixel Complex Scenes (CVPR 2024 · similarity 0.86)
- SocialCircle: Learning the Angle-based Social Interaction Representation for Pedestrian Trajectory Prediction (CVPR 2024 · 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.