Research map / Point clouds / Point cloud classification
GSOT3D: Towards Generic 3D Single Object Tracking in the Wild
Yifan Jiao, Yunhao Li, Junhua Ding, Qing Yang, Song Fu, Heng Fan, Libo 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 |
|---|---|---|---|
| add 9DoF box prediction instead of 7DoF (ablation) | PROT3D · GSOT3D test | mAO (%): 19.86 -> 20.03 | Improved |
| add 9DoF box prediction instead of 7DoF (ablation) | PROT3D · GSOT3D test | mSR75 (%): 2.36 -> 3.29 | Improved |
| add 9DoF box prediction instead of 7DoF (ablation) | PROT3D · GSOT3D test | mSR50 (%): 15.16 -> 15.46 | Improved |
| add progressive architecture (two stages) (ablation) | PROT3D · GSOT3D test | mAO (%): 20.03 -> 21.97 | Improved |
| add progressive architecture (two stages) (ablation) | PROT3D · GSOT3D test | mSR50 (%): 15.46 -> 19.76 | Improved |
| add progressive architecture (two stages) (ablation) | PROT3D · GSOT3D test | mSR75 (%): 3.29 -> 5.22 | Improved |
| use 3 stages instead of 2 (ablation) | PROT3D · GSOT3D test | mAO (%): 21.97 -> 21.58 | Hurt |
| use 3 stages instead of 2 (ablation) | PROT3D · GSOT3D test | mSR50 (%): 19.76 -> 19.61 | Hurt |
| use memory size K=2 instead of 3 (ablation) | PROT3D · GSOT3D test | mAO (%): 21.97 -> 21.37 | Hurt |
| use memory size K=4 instead of 3 (ablation) | PROT3D · GSOT3D test | mAO (%): 21.97 -> 21.84 | Hurt |
| retrain trackers on GSOT3D training set instead of KITTI-pretrained | MBPTrack · GSOT3D test | mAO (%): 3.38 -> 20.54 | Improved |
| retrain trackers on GSOT3D training set instead of KITTI-pretrained | M2-Track · GSOT3D test | mAO (%): 3.65 -> 20.26 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- M3SOT — M3SOT: Multi-Frame, Multi-Field, Multi-Space 3D Single Object Tracking (AAAI 2024) · better on 3 of 3 shared results
- CXTrack — CXTrack: Improving 3D Point Cloud Tracking With Contextual Information (CVPR 2023) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- M3SOT: Multi-Frame, Multi-Field, Multi-Space 3D Single Object Tracking (AAAI 2024 · similarity 0.90)
- Towards Generic 3D Tracking in RGBD Videos: Benchmark and Baseline (ECCV 2022 · similarity 0.90)
- TAPVid-3D: A Benchmark for Tracking Any Point in 3D (NeurIPS 2024 · similarity 0.88)
- TrackingWorld: World-centric Monocular 3D Tracking of Almost All Pixels (NeurIPS 2025 · similarity 0.87)
- TAPIP3D: Tracking Any Point in Persistent 3D Geometry (NeurIPS 2025 · similarity 0.87)
- TrackCraft3R: Repurposing Video Diffusion Transformers for Dense 3D Tracking (NeurIPS 2026 · similarity 0.87)
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.
