Research map / Multi-agent systems and games / Human-AI and goal-conditioned
DESIRE: Distant Future Prediction in Dynamic Scenes With Interacting Agents
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 IOC ranking, regression, and iterative refinement (ablation) | DESIRE-SI on KITTI · KITTI, 4.0 sec | L2 error (m): 2.72 -> 2.06 | Helped |
| add IOC ranking, regression, and iterative refinement (ablation) | DESIRE-SI on SDD · SDD, 4.0 sec | pixel error: 7.61 -> 5.33 | Helped |
| add agent interaction features (ablation) | DESIRE-S-IT4 · KITTI, 4.0 sec | L2 error (m): 2.07 -> 2.06 | Hurt |
| add agent interaction features (ablation) | DESIRE-S-IT4 · SDD, 4.0 sec | pixel error: 5.62 -> 5.33 | Helped |
| use multiple ranked predictions rather than one best prediction (ablation) | DESIRE-SI-IT4 · SDD, 4.0 sec | pixel error: 10.65 -> 5.33 | Helped |
| increase the number of prediction samples from K=25 to K=200 (ablation) | DESIRE-S-IT4 top 20 · KITTI, 4.0 sec | oracle L2 error (m), top 20: 2.03 -> 1.96 | Helped |
| increase the number of prediction samples from K=25 to K=200 (ablation) | DESIRE-S-IT4 Best · KITTI, 4.0 sec | L2 error (m): 4.87 -> 4.70 | No clear effect |
| shorten observed past trajectory from 2.0 sec to 1.0 sec (ablation) | DESIRE-S-IT4 Best · KITTI, 4.0 sec | L2 error (m): 4.71 -> 4.94 | Hurt |
| use RNN encoder-decoder instead of a linear regressor | RNN ED · KITTI, 4.0 sec | L2 error (m): 5.62 -> 3.86 | Helped |
| add scene and interaction features to RNN ED | RNN ED on KITTI · KITTI, 4.0 sec | L2 error (m): 3.86 -> 4.29 | Hurt |
| add scene and interaction features to RNN ED | RNN ED on SDD · SDD, 4.0 sec | pixel error: 9.54 -> 8.80 | Helped |
| generate multiple CVAE hypotheses instead of one | CVAE on KITTI · KITTI, 4.0 sec | L2 error (m): 6.16 -> 3.07 | Helped |
Closest papers by meaning
- Advancing Semantic Future Prediction through Multimodal Visual Sequence Transformers (CVPR 2025 · similarity 0.87)
- SAMPO: Scale-wise Autoregression with Motion Prompt for Generative World Models (NeurIPS 2025 · similarity 0.87)
- Multi-scale Temporal Prediction via Incremental Generation and Multi-agent Collaboration (NeurIPS 2025 · similarity 0.86)
- MESSENGER: Memory-Enhanced Sequential Scene Flow Estimation via Autoregressive Next-Frame Forecasting (NeurIPS 2026 · similarity 0.86)
- ThinkJEPA: Empowering Latent World Models with Large Vision-Language Reasoning Model (NeurIPS 2026 · similarity 0.85)
- PlaySlot: Learning Inverse Latent Dynamics for Controllable Object-Centric Video Prediction and Planning (ICML 2025 · 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.