Research map / Detection and segmentation / 3D detection and driving
You Don't Only Look Once: Constructing Spatial-Temporal Memory for Integrated 3D Object Detection and Tracking
Jiaming Sun, Yiming Xie, Siyu Zhang, Linghao Chen, Guofeng Zhang, Hujun Bao, Xiaowei Zhou
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 middle integration (Fusion R-CNN with back-end predictions) (ablation) | UDOLO · ScanNet val (depth, all objects) | mAP@0.5 IoU: 14.78 -> 18.43 | Improved |
| add middle integration (ablation) | UDOLO · ScanNet val chair (depth) | AP@0.5 IoU: 41.17 -> 51.63 | Improved |
| add early integration (OOM-guided RPN) (ablation) | UDOLO · ScanNet val (depth, all objects) | mAP@0.5 IoU: 18.26 -> 18.43 | No clear effect |
| add early integration (OOM-guided RPN) (ablation) | UDOLO with PointRCNN · KITTI Object | backbone+RPN runtime: 42ms -> 32ms | Improved |
| add spatial-temporal memory feedback to single-frame VoteNet | VoteNet · ScanNet val (depth, occluded) | mAP@0.5 IoU: 0.91 -> 3.23 | Improved |
| add spatial-temporal memory feedback to single-frame VoteNet | VoteNet · ScanNet val (depth, all objects) | mAP@0.5 IoU: 14.75 -> 18.43 | Improved |
| add spatial-temporal memory feedback to single-frame VoteNet | VoteNet · ScanNet val (RGB-reconstructed, all objects) | mAP@0.5 IoU: 4.41 -> 7.32 | Improved |
| add spatial-temporal memory feedback | VoteNet · ScanNet bounding box stability | Std. rotation: 37.98 -> 17.88 | Improved |
| add spatial-temporal memory feedback to AB3DMOT detector | AB3DMOT (PointRCNN) · KITTI Tracking val car, LiDAR, 3D IoU=0.7 | sAMOTA: 69.8 -> 75.7 | Improved |
| add spatial-temporal memory feedback to stereo tracker | JST · KITTI Tracking val car, stereo, 3D IoU=0.5 | MOTA: 29.4 -> 45.5 | Improved |
| add spatial-temporal memory feedback to stereo tracker | JST · KITTI Tracking val car, stereo, 3D IoU=0.25 | MOTA: 56.7 -> 53.7 | Hurt |
| add spatial-temporal memory feedback to PL(PointRCNN) | PointRCNN · KITTI Object val car, AP3D IoU=0.7 Easy | AP3D: 62.51 -> 70.27 | Improved |
Closest papers by meaning
- MAD: Memory-Augmented Detection of 3D Objects (CVPR 2025 · similarity 0.89)
- Video Object Detection with an Aligned Spatial-Temporal Memory (ECCV 2018 · similarity 0.88)
- MeMOT: Multi-Object Tracking With Memory (CVPR 2022 · similarity 0.88)
- Time3D: End-to-End Joint Monocular 3D Object Detection and Tracking for Autonomous Driving (CVPR 2022 · similarity 0.88)
- PTT: Point-Trajectory Transformer for Efficient Temporal 3D Object Detection (CVPR 2024 · similarity 0.86)
- A Distractor-Aware Memory for Visual Object Tracking with SAM2 (CVPR 2025 · 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.
