Research map / Detection and segmentation / Tracking
Instance-level Visual Active Tracking with Occlusion-Aware Planning
Haowei Sun, Kai Zhou, Hao Gao, Shiteng Zhang, Jinwu Hu, Xutao Wen, Qixiang Ye, Mingkui Tan
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-view instance-aware prototype instead of raw DINOv3 features (ablation) | OA-VAT · UnrealCV with distractors | avg success rate: 0.87 -> 0.93 | Improved |
| add online prototype enhancement (EMA) (ablation) | OA-VAT · UnrealCV with distractors | avg success rate: 0.82 -> 0.93 (+13.4% avg. SR) | Improved |
| use EMA instead of average update for prototype enhancement (ablation) | OA-VAT · UnrealCV with distractors | avg success rate: 0.89 -> 0.93 (+4.5% avg. SR) | Improved |
| add confidence-aware Kalman filter (ablation) | OA-VAT · UnrealCV with distractors | avg success rate: 0.87 -> 0.93 | Improved |
| make Kalman filter confidence-aware instead of linear (ablation) | OA-VAT · UnrealCV with distractors | avg success rate: 0.90 -> 0.93 (+3.3% avg. SR) | Improved |
| add diffusion trajectory planner instead of PID controller (ablation) | OA-VAT · UnrealCV with distractors | avg success rate: 0.85 -> 0.93 | Improved |
| use proposed planner instead of EVT offline-RL planning (ablation) | OA-VAT · UnrealCV with distractors | avg success rate: 0.87 -> 0.93 (+6.9%) | Improved |
| condition planner on target bounding box (ablation) | OA-VAT planner · UnrealCV with distractors | avg success rate: 0.89 -> 0.93 (+4.5%) | Improved |
| raise EMA momentum from 0.8 to 0.9 (ablation) | OA-VAT · UnrealCV with distractors | avg success rate: 0.93 -> 0.90 | Hurt |
| lower EMA momentum from 0.8 to 0.6 (ablation) | OA-VAT · UnrealCV with distractors | avg success rate: 0.93 -> 0.93 | No clear effect |
| vary Kalman filter slope lambda from 15 to 13 or 17 (ablation) | OA-VAT · UnrealCV with distractors | avg success rate: 0.93 -> 0.91 (13), 0.93 (17) | No clear effect |
| vary Kalman filter center gamma from 0.4 to 0.3 or 0.5 (ablation) | OA-VAT · UnrealCV with distractors | avg success rate: 0.93 -> 0.92 (0.3), 0.93 (0.5) | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- RSPT — RSPT: Reconstruct Surroundings and Predict Trajectory for Generalizable Active Object Tracking (AAAI 2023) · better on 1 of 1 shared result
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
- Open-World Drone Active Tracking with Goal-Centered Rewards (NeurIPS 2025 · similarity 0.88)
- Stabilizing Real-World Visual Active Tracking with Action-Smooth Test-Time Adaptation (ECCV 2026 · similarity 0.88)
- Learning Occlusion-Robust Vision Transformers for Real-Time UAV Tracking (CVPR 2025 · similarity 0.88)
- Learning Adaptive and View-Invariant Vision Transformer for Real-Time UAV Tracking (ICML 2024 · similarity 0.86)
- LEAP-VO: Long-term Effective Any Point Tracking for Visual Odometry (CVPR 2024 · similarity 0.85)
- Instance Tracking in 3D Scenes from Egocentric Videos (CVPR 2024 · similarity 0.84)
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.
