Research map / Detection and segmentation / Tracking
FAMNet: Joint Learning of Feature, Affinity and Multi-Dimensional Assignment for Online Multiple Object Tracking
Peng Chu, Haibin Ling
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 |
|---|---|---|---|
| fine-tune from ILSVRC15 pretrained weights instead of training from scratch (ablation) | FAMNet · MOT2015 ablation sequences | MOTA: 44.1 -> 45.2 | Improved |
| train the network on MOT sequence instead of random initialization (ablation) | FAMNet · MOT2015 ablation sequences | MOTA: 35.5 -> 44.1 | Improved |
| add dedicated CNN_BBE bounding box quality estimator (ablation) | FAMNet · MOT2015 ablation sequences | MOTA: 40.5 -> 45.2 | Improved |
| add CNN_BBE bounding box quality estimator (ablation) | FAMNet · MOT2015 ablation sequences | FP: 518 -> 259 | Improved |
| add SOT predictions via virtual candidates (ablation) | FAMNet · MOT2015 ablation sequences | MOTA: 42.0 -> 45.2 | Improved |
| add SOT predictions via virtual candidates (ablation) | FAMNet · MOT2015 ablation sequences | FN: 4412 -> 4105 | Improved |
| add SOT predictions via virtual candidates (ablation) | FAMNet · MOT2015 ablation sequences | FP: 200 -> 259 | Hurt |
| add SOT predictions via virtual candidates (ablation) | FAMNet · MOT2015 ablation sequences | IDS: 99 -> 87 | Improved |
| use end-to-end learned higher-order affinity instead of RNN end-to-end online trackers | FAMNet · MOT2015 test | MOTA: 19.0 -> 40.6 | Improved |
| use FAMNet instead of hand-crafted R1TA | FAMNet · KITTI-Car test | MOTA: 71.2 -> 77.1 | Improved |
| use FAMNet instead of best online baseline | FAMNet · MOT2017 test | MOTA: 50.9 -> 52.0 | Improved |
| use FAMNet instead of best offline method | FAMNet · KITTI-Car test | MOTA: 78.1 -> 77.1 | Hurt |
Closest papers by meaning
- A Unified Object Motion and Affinity Model for Online Multi-Object Tracking (CVPR 2020 · similarity 0.90)
- Deep Affinity Network for Multiple Object Tracking (TPAMI 2019 · similarity 0.90)
- DASOT: A Unified Framework Integrating Data Association and Single Object Tracking for Online Multi-Object Tracking (AAAI 2020 · similarity 0.88)
- Focusing on Tracks for Online Multi-Object Tracking (CVPR 2025 · similarity 0.86)
- Multiple Object Tracking as ID Prediction (CVPR 2025 · similarity 0.86)
- DeconfuseTrack: Dealing with Confusion for Multi-Object Tracking (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.
