Research map / Probabilistic and scientific modeling / Molecules and proteins
StretchBEV: Stretching Future Instance Prediction Spatially and Temporally
Adil Kaan Akan, Fatma Güney
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 unsupervised pre-training of the dynamics (ablation) | StretchBEV · nuScenes short (near) | IoU: 53.3 -> 55.5 | Improved |
| add unsupervised pre-training of the dynamics (ablation) | StretchBEV · nuScenes short (far) | VPQ: 26.0 -> 29.0 | Improved |
| add pre-training of the dynamics (ablation) | StretchBEV · nuScenes short (near) | VPQ: 41.7 -> 46.0 | Improved |
| add output-modality labels to the posterior (StretchBEV-P vs StretchBEV) (ablation) | StretchBEV · nuScenes short (far) | VPQ: 26.0 -> 47.5 | Improved |
| add output-modality labels to the posterior (ablation) | StretchBEV · nuScenes short (far) | IoU: 35.8 -> 52.5 | Improved |
| replace FIERY with StretchBEV-P (per-step stochastic residual latents) | FIERY · nuScenes short (far) | VPQ: 29.0 -> 47.5 | Improved |
| replace FIERY with StretchBEV-P | FIERY · nuScenes short (near) | IoU: 58.8 -> 58.1 | Hurt |
| add content variable (ablation) | StretchBEV · nuScenes short (near) | IoU: 53.3 -> 51.9 | Hurt |
| add content variable (ablation) | StretchBEV-P · nuScenes short (far) | VPQ: 47.5 -> 46.8 | Hurt |
| replace FIERY with StretchBEV-P over long horizon (far VPQ) | FIERY · nuScenes long horizon (far) | VPQ: 20.9 -> 30.8 | Improved |
| replace FIERY with StretchBEV-P over long horizon (near IoU) | FIERY · nuScenes long horizon (near) | IoU: 41.8 -> 38.2 | Hurt |
| replace FIERY with StretchBEV for diversity (long far) | FIERY · nuScenes diversity | Generalized Energy Distance: 152.38 -> 109.12 | Improved |
Closest papers by meaning
- Motion Forecasting in Continuous Driving (NeurIPS 2024 · similarity 0.86)
- FutureSightDrive: Thinking Visually with Spatio-Temporal CoT for Autonomous Driving (NeurIPS 2025 · similarity 0.86)
- Future-Aware End-to-End Driving: Bidirectional Modeling of Trajectory Planning and Scene Evolution (NeurIPS 2025 · similarity 0.86)
- DINO-Foresight: Looking into the Future with DINO (NeurIPS 2025 · similarity 0.86)
- DriveFuture: Future-Aware Latent World Models for Autonomous Driving (NeurIPS 2026 · similarity 0.85)
- Visual Point Cloud Forecasting enables Scalable Autonomous Driving (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.
