Research map / 3D vision / 3D generation
SimAug: Learning Robust Representations from Simulation for Trajectory Prediction
Junwei Liang, Lu Jiang, Alexander Hauptmann
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 top-down view data to multi-view simulation (ablation) | SimAug · Stanford Drone | minADE1: 18.4 -> 15.7 | Improved |
| add top-down view data to multi-view simulation (ablation) | SimAug · VIRAT/ActEV | minADE1: 22.8 -> 21.7 | Improved |
| add random perturbation (ablation) | SimAug · Argoverse | minADE1: 69.1 -> 67.9 | Improved |
| add random perturbation (ablation) | SimAug · Stanford Drone | minADE1: 18.7 -> 15.7 | Improved |
| add adversarial attack (ablation) | SimAug · Stanford Drone | minADE1: 17.4 -> 15.7 | Improved |
| add adversarial attack (ablation) | SimAug · VIRAT/ActEV | minADE1: 23.1 -> 21.7 | Improved |
| add hardest-view selection instead of random search (ablation) | SimAug · Stanford Drone | minADE1: 19.6 -> 15.7 | Improved |
| add hardest-view selection instead of random search (ablation) | SimAug · Argoverse | minADE1: 68.6 -> 67.9 | Improved |
| use SimAug instead of standard augmentation | Multiverse · Stanford Drone | Grid_Acc: 21.3 -> 45.4 | Improved |
| use SimAug instead of FGSM | Multiverse · Argoverse | minADE1: 71.1 -> 67.9 | Improved |
| use PGD adversarial augmentation | Multiverse · Argoverse | minADE1: 69.1 -> 72.8 | Hurt |
| fine-tune SimAug on real training videos | SimAug · Stanford Drone (K=20) | minADE20: 12.03 -> 10.27 | Improved |
Closest papers by meaning
- Certified Human Trajectory Prediction (CVPR 2025 · similarity 0.83)
- Learning Occlusion-Robust Vision Transformers for Real-Time UAV Tracking (CVPR 2025 · similarity 0.83)
- CaDeT: a Causal Disentanglement Approach for Robust Trajectory Prediction in Autonomous Driving (CVPR 2024 · similarity 0.83)
- Physical Plausibility-aware Trajectory Prediction via Locomotion Embodiment (CVPR 2025 · similarity 0.82)
- FutureHuman3D: Forecasting Complex Long-Term 3D Human Behavior from Video Observations (CVPR 2024 · similarity 0.82)
- WorldVLN: Autoregressive World Action Model for Aerial Vision-Language Navigation (NeurIPS 2026 · similarity 0.82)
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.
