Research map / Detection and segmentation / Tracking
Tracking Occluded Objects and Recovering Incomplete Trajectories by Reasoning About Containment Relations and Human Actions
Wei Liang, Yixin Zhu, Song-Chun Zhu
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 human-action temporal term (transition probabilities) (ablation) | Full model · Occluded object dataset, all | success rate (r0=0.5): 0.57 -> 0.69 | Improved |
| add detection-based spatial term (ablation) | Full model · Occluded object dataset, all | success rate (r0=0.5): 0.32 -> 0.69 | Improved |
| add human-action temporal term, contained subset (ablation) | Full model · Occluded object dataset, contained | success rate (r0=0.5): 0.15 -> 0.42 | Improved |
| add human-action temporal term, blocked subset (ablation) | Full model · Occluded object dataset, blocked | success rate (r0=0.5): 0.21 -> 0.47 | Improved |
| full model vs explicit occlusion model baseline | Full model · Occluded object dataset, all | success rate (r0=0.5): 0.59 -> 0.69 | Improved |
| full model vs baseline3 on contained subset | Full model · Occluded object dataset, contained | success rate (r0=0.5): 0.16 -> 0.42 | Improved |
| full model vs baseline3 on CAD-120 | Full model · CAD-120 | success rate (r0=0.5): 0.33 -> 0.47 | Improved |
| full model vs baseline3 on NW-UCLA | Full model · NW-UCLA | success rate (r0=0.5): 0.56 -> 0.72 | Improved |
| add object context and global DP optimization to human-pose-only HOI classifier (ablation) | HOI consistency · Occluded object dataset HOI | confusion matrix | Improved |
Closest papers by meaning
- Tracking Through Containers and Occluders in the Wild (CVPR 2023 · similarity 0.88)
- Tracking Without Bells and Whistles (ICCV 2019 · similarity 0.86)
- On Occlusions in Video Action Detection: Benchmark Datasets And Training Recipes (NeurIPS 2023 · similarity 0.86)
- Learning To Track With Object Permanence (ICCV 2021 · similarity 0.86)
- Detecting Invisible People (ICCV 2021 · similarity 0.86)
- More Than Meets the Eye: Enhancing Multi-Object Tracking Even with Prolonged Occlusions (ICML 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.
