Research map / Video / Video segmentation and tracking
Connecting the Complementary-View Videos: Joint Camera Identification and Subject Association
Ruize Han, Yiyang Gan, Jiacheng Li, Feifan Wang, Wei Feng, Song Wang
Check 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 spatial-aware position representations with joint camera identification and subject association | Complementary-view video analysis · Synthetic dataset, camera location | AUC: 69.95 -> 80.50 | Improved |
| Use spatial-aware position representations with joint camera identification and subject association | Complementary-view video analysis · Real-world dataset, camera location | AUC: 60.61 -> 79.26 | Improved |
| Use spatial-aware position representations with joint camera identification and subject association | Complementary-view video analysis · Synthetic dataset, subject association | F1 score: 45.09 -> 68.16 | Improved |
| Add direction loss LD (ablation) | Camera identification module · Synthetic dataset, camera location | AUC: 73.72 -> 80.50 | Improved |
| Add localization loss LL (ablation) | Camera identification module · Synthetic dataset, camera location | AUC: 72.69 -> 80.50 | Improved |
| Add heatmap compression before correlation (ablation) | Camera identification module · Synthetic dataset, camera location | AUC: 71.51 -> 80.50 | Improved |
| Add heatmap compression before correlation (ablation) | Camera identification module · Synthetic dataset, view direction | AUC: 60.29 -> 68.34 | Improved |
| Combine x-axis and y-axis matching similarities (ablation) | Subject association module · Synthetic dataset, subject association | F1 score: 59.24 -> 68.16; 40.73 -> 68.16 | Improved |
| Combine x-axis and y-axis matching similarities (ablation) | Subject association module · Real-world dataset, subject association | F1 score: 74.04 -> 76.81; 43.80 -> 76.81 | Improved |
| Use bounding-box height rather than estimated depth or bottom distance for y-axis distribution (ablation) | Subject association module · Synthetic dataset, subject association | F1 score: 60.46 -> 68.16; 64.32 -> 68.16 | Improved |
| Use bounding-box height rather than estimated depth or bottom distance for y-axis distribution (ablation) | Subject association module · Real-world dataset, subject association | F1 score: 74.90 -> 76.81; 74.82 -> 76.81 | Improved |
| Train on synthetic data and test on real-world data without adaptation | Ours · Real-world dataset, cross-domain subject association | F1 score: 57.85 | Mixed |
Closest papers by meaning
- Multiple Human Association and Tracking from Egocentric and Complementary Top Views (TPAMI 2021 · similarity 0.90)
- Complementary-View Multiple Human Tracking (AAAI 2020 · similarity 0.88)
- Egocentric Meets Top-View (TPAMI 2018 · similarity 0.86)
- Integrating Egocentric Videos in Top-view Surveillance Videos: Joint Identification and Temporal Alignment (ECCV 2018 · similarity 0.85)
- View From Above: Orthogonal-View aware Cross-view Localization (CVPR 2024 · similarity 0.82)
- Drones Help Drones: A Collaborative Framework for Multi-Drone Object Trajectory Prediction and Beyond (NeurIPS 2024 · similarity 0.82)
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.
