Research map / Probabilistic and scientific modeling / Molecules and proteins
SEEM: A Sequence Entropy Energy-Based Model for Pedestrian Trajectory All-Then-One Prediction
Dafeng Wang, Hongbo Liu, Naiyao Wang, Yiyang Wang, Hua Wang, Seán McLoone
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 |
|---|---|---|---|
| Add probability distribution clipping (PDC) module (ablation) | SEEM · ETH/UCY (100% training data, t_pred=12) | ADE: 10% to 26% lower ADE | Improved |
| Add probability distribution clipping (PDC) module (ablation) | SEEM · ETH/UCY (100% training data, t_pred=12) | FDE: 11% to 21% lower FDE | Improved |
| Increase PDC steps from 2 to 4 (ablation) | SEEM · ETH/UCY | ADE/FDE | Improved |
| Increase PDC steps beyond 4 up to 22 (ablation) | SEEM · ETH/UCY | ADE/FDE | No clear effect |
| Increase LSTM hidden size in GeneratorSO from 64 to 128-512 (ablation) | SEEM · ETH/UCY | ADE/FDE | Hurt |
| Set diversity/accuracy loss weight lambda to 0.5 instead of 0 or 1 (ablation) | SEEM · ETH/UCY | ADE/FDE | Improved |
| Add mutual information loss L_mi (ablation) | SEEM · ETH/UCY (t_pred=8) | ADE: 15.6% better ADE | Improved |
| Add mutual information loss L_mi (ablation) | SEEM · ETH/UCY (t_pred=12) | FDE: 22.0% and 38.3% better FDE | Improved |
| Use SEEM instead of LB-EBM | LB-EBM · ETH/UCY average, t_pred=12 | FDE (m): 1.08 -> 0.95 | Improved |
| Use SEEM instead of LB-EBM | LB-EBM · ETH/UCY average, t_pred=8 | ADE (m): 0.39 -> 0.38 | Improved |
| Use SEEM instead of Trajectron++ | Trajectron++ · ETH/UCY average | KDE NLL: -0.74 -> -0.96 | Improved |
| Use SEEM instead of Trajectron++ | Trajectron++ · TrajNet++ (K=3) | TopK FDE (m): 1.05 -> 0.95 | Improved |
Closest papers by meaning
- Non-Probability Sampling Network for Stochastic Human Trajectory Prediction (CVPR 2022 · similarity 0.88)
- BCDiff: Bidirectional Consistent Diffusion for Instantaneous Trajectory Prediction (NeurIPS 2023 · similarity 0.88)
- Pedestrian Trajectory Prediction with Missing Data: Datasets, Imputation, and Benchmarking (NeurIPS 2024 · similarity 0.86)
- SocialCVAE: Predicting Pedestrian Trajectory via Interaction Conditioned Latents (AAAI 2024 · similarity 0.86)
- Towards Predicting Any Human Trajectory In Context (NeurIPS 2025 · similarity 0.86)
- TrajCLIP: Pedestrian trajectory prediction method using contrastive learning and idempotent networks (NeurIPS 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.
