Research map / 3D vision / 3D generation
GOT-Edit: Geometry-Aware Generic Object Tracking via Online Model Editing
Shih-Fang Chen, Jun-Cheng Chen, I-hong Jhuo, Yen-Yu Lin
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 VGGT geometry features with null-space constrained online model editing | ToMP-378 · AVisT/NfS/LaSOT | SUC: AVisT 61.5 -> 63.7 (Base), LaSOT 72.6 -> 75.3 | Improved |
| use geometry features alone (VGGT DPT) instead of semantic (ablation) | GOT-Edit-252 · AVisT | SUC: 59.2 -> 55.8 | Hurt |
| naive fusion of geometry with semantics (no null-space) (ablation) | GOT-Edit-252 · LaSOT | SUC: 70.7 -> 71.3 | Improved |
| naive fusion of geometry with semantics (ablation) | GOT-Edit-252 · LaSOT Fast Motion / AVisT Distractor | SUC: 60.39 -> 58.73; 49.38 -> 47.18 | Hurt |
| naive fusion of geometry with semantics (ablation) | GOT-Edit-252 · LaSOT Partial Occlusion | SUC: 68.97 -> 70.08 | Improved |
| add null-space constraint to fusion (ablation) | GOT-Edit-252 · LaSOT | SUC: 71.3 -> 72.7 | Improved |
| add whitening and ridge regularization before SVD (ablation) | GOT-Edit-252 · LaSOT | SUC: 72.7 -> 73.8 | Improved |
| use independent DINO backbone instead of VGGT's DINO for semantics (ablation) | GOT-Edit-252 · NfS | SUC: 67.5 -> 68.5 | Improved |
| add VastTrack training data | GOT-Edit-378 · AVisT | SUC: 63.7 -> 64.5 | Improved |
| add VastTrack training data | GOT-Edit-378 · LaSOT | SUC: 75.3 -> 75.0 | No clear effect |
| replace VGGT by StreamVGGT with memory cache 1 (ablation) | GOT-Edit-252 · LaSOT | SUC: 73.8 -> 72.8 (LaSOT); runtime 84.1 -> 72.5 ms | Mixed |
Closest papers by meaning
- G$^2$TAM: Geometry Grounded Track Anything Model (ICML 2026 · similarity 0.89)
- Motion-Aware Object Tracking via Motion and Geometry-Aware Cues (AAAI 2026 · similarity 0.87)
- RegTrack: Simplicity Beneath Complexity in Robust Multi-Modal 3D Multi-Object Tracking (TPAMI 2026 · similarity 0.87)
- Fast Spatial Tracking with Visual Geometry Transformer (CVPR 2026 · similarity 0.86)
- Tracking by 3D Model Estimation of Unknown Objects in Videos (ICCV 2023 · similarity 0.86)
- Prior-free 3D Object Tracking (CVPR 2025 · 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.
