Research map / 3D vision / Human pose estimation
Learning to Detect and Track Visible and Occluded Body Joints in a Virtual World
Matteo Fabbri, Fabio Lanzi, Simone Calderara, Andrea Palazzi, Roberto Vezzani, Rita Cucchiara
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 occlusion branch (ablation) | Single Image model · JTA test (detection) | joints mAP: 50.9 -> 56.3 | Improved |
| add occlusion branch (ablation) | Single Image model · JTA test (detection) | detection F1: 71.6 -> 78.4 | Improved |
| add temporal branch with TAFs (ablation) | Single Image + occ · JTA test (detection) | joints mAP: 56.3 -> 59.3 | Improved |
| add temporal branch with TAFs (ablation) | Single Image + occ · JTA test (detection) | detection F1: 78.4 -> 83.9 | Improved |
| add temporal branch with TAFs (ablation) | Single Image + occ · JTA test (detection) | detection recall: 71.8 -> 77.4 | Improved |
| distance-scaled Gaussian heatmaps instead of fixed scale | prior pose network · JTA test (detection) | joints mAP: 50.1 -> 50.9 | Improved |
| use THOPA-net detections instead of DPM | Hungarian tracker · JTA test (tracking) | MOTA: 31.5 -> 57.4 | Improved |
| use TAF-based association instead of Hungarian on bounding box midpoint | THOPA-net · JTA test (tracking) | MOTA: 57.4 -> 59.3 | Improved |
| use TAF-based association instead of Hungarian on bounding box midpoint | THOPA-net · JTA test (tracking) | IDF1: 57.3 -> 63.2 | Improved |
| use TAF-based association | THOPA-net · JTA test (tracking) | ID switches: 15236 -> 10214 | Improved |
| fine-tune on real data w/o appearance model | THOPA-net · MOT-16 test (online trackers) | IDF1: 29.2 | Hurt |
Closest papers by meaning
- Multi-Person Articulated Tracking With Spatial and Temporal Embeddings (CVPR 2019 · similarity 0.87)
- ArtTrack: Articulated Multi-Person Tracking in the Wild (CVPR 2017 · similarity 0.86)
- DeepCut: Joint Subset Partition and Labeling for Multi Person Pose Estimation (CVPR 2016 · similarity 0.86)
- Efficient Online Multi-Person 2D Pose Tracking With Recurrent Spatio-Temporal Affinity Fields (CVPR 2019 · similarity 0.86)
- OpenPose: Realtime Multi-Person 2D Pose Estimation Using Part Affinity Fields (TPAMI 2019 · similarity 0.86)
- CoMotion: Concurrent Multi-person 3D Motion (ICLR 2025 · 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.
