Research map / Label-efficient and robust learning / Recognition and long-tail
FipTR: A Simple yet Effective Transformer Framework for Future Instance Prediction in Autonomous Driving
Xingtai Gui, Tengteng Huang, Haonan Shao, Haotian Yao, Chi Zhang
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 |
|---|---|---|---|
| use multi-frame mask matching cost instead of single-frame mask cost in Hungarian matching (ablation) | FipTR-T · nuScenes val | VPQ(n): 51.1 -> 53.3 | Improved |
| add mask cost to the matching cost (ablation) | FipTR-T · nuScenes val | VPQ(n): 48.2 -> 53.3 | Improved |
| add the auxiliary 3D box branch to the instance mask decoder (ablation) | FipTR-T · nuScenes val | IoU / VPQ(n): 5.2 -> 59.0 IoU, 1.1 -> 33.6 VPQ (VPQ(f) column) | Improved |
| add flow-aware deformable attention in place of vanilla deformable attention (ablation) | FipTR-T · nuScenes val | VPQ(n): 52.5 -> 53.3 | Improved |
| add FADA at far range (100m x 100m ROI) (ablation) | FipTR-S · nuScenes val | IoU(f): 40.4 -> 40.4 | No clear effect |
| replace RNN future BEV module with deformable attention based module (ablation) | FipTR-T · nuScenes val | VPQ(n): 46.7 -> 52.5 | Improved |
| use non-weight-shared Context CNN across future frames (ablation) | FipTR-T · nuScenes val | VPQ(n): 41.9 -> 53.3 | Improved |
| use non-weight-shared time-specific instance queries (ablation) | FipTR-T · nuScenes val | VPQ(n): 52.5 -> 53.3 | Improved |
| replace BEVerse with end-to-end FipTR on BEVDet4D encoder (Swin-S) | BEVerse-S · nuScenes val | VPQ(n): 51.7 -> 56.1 | Improved |
| replace UniAD with FipTR-R101 on BEVFormer encoder | UniAD · nuScenes val | IoU(n): 63.4 -> 62.9 | Mixed |
| replace PowerBEV with FipTR-T on IoU(n) | PowerBEV · nuScenes val | IoU(n): 62.5 -> 59.0 | Hurt |
| add segmentation (mask) branch alongside detection (ablation) | FipTR-T · nuScenes val | mAVE: 0.78 -> 0.51 | Improved |
Closest papers by meaning
- Think Twice Before Driving: Towards Scalable Decoders for End-to-End Autonomous Driving (CVPR 2023 · similarity 0.87)
- DriveTransformer: Unified Transformer for Scalable End-to-End Autonomous Driving (ICLR 2025 · similarity 0.85)
- Future-Aware End-to-End Driving: Bidirectional Modeling of Trajectory Planning and Scene Evolution (NeurIPS 2025 · similarity 0.85)
- M-BEV: Masked BEV Perception for Robust Autonomous Driving (AAAI 2024 · similarity 0.84)
- Visual Point Cloud Forecasting enables Scalable Autonomous Driving (CVPR 2024 · similarity 0.83)
- SmartPretrain: Model-Agnostic and Dataset-Agnostic Representation Learning for Motion Prediction (ICLR 2025 · similarity 0.83)
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.
