Research map / Detection and segmentation / Tracking
MOTS: Multi-Object Tracking and Segmentation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| train jointly on temporally consistent mask annotations with temporal convolutions and learned association | TrackR-CNN · KITTI MOTS val, Car | sMOTSA: 75.1 -> 76.2 | Helped |
| train jointly on temporally consistent mask annotations with temporal convolutions and learned association | TrackR-CNN · KITTI MOTS val, Ped | sMOTSA: 45.0 -> 46.8 | Helped |
| use TrackR-CNN instead of MOTChallenge bounding-box trackers with generated masks | MOTSChallenge, Ped · MOTSChallenge | sMOTSA: 49.3 -> 52.7 | Helped |
| add two 3D convolution layers rather than no temporal component (ablation) | TrackR-CNN · KITTI MOTS val, Ped | sMOTSA: 44.8 -> 46.8 | Helped |
| add two 3D convolution layers rather than no temporal component (ablation) | TrackR-CNN · KITTI MOTS val, Car | sMOTSA: 76.4 -> 76.2 | No clear effect |
| replace temporal convolutions with ConvLSTM layers (ablation) | TrackR-CNN · KITTI MOTS val | sMOTSA: 76.4 vs 75.7; 44.8 vs 45.0 | No clear effect |
| add a learned association head rather than use mask propagation scores (ablation) | TrackR-CNN · KITTI MOTS val, Car | sMOTSA: 75.5 -> 76.2 | Helped |
| add a learned association head rather than use mask propagation scores (ablation) | TrackR-CNN · KITTI MOTS val, Ped | sMOTSA: 46.1 -> 46.8 | Helped |
| add association-loss training while using mask IoU for matching (ablation) | TrackR-CNN · KITTI MOTS val, Ped | sMOTSA: 44.9 -> 46.1 | Helped |
| use bounding-box center distance instead of learned association vectors (ablation) | TrackR-CNN · KITTI MOTS val, Ped | sMOTSA: 46.8 -> 43.3 | Hurt |
| use batch-hard triplet loss instead of batch-all triplet loss (ablation) | TrackR-CNN association head · KITTI MOTS val, Ped | sMOTSA: 45.3 -> 46.8 | Helped |
| use batch-hard triplet loss instead of contrastive loss (ablation) | TrackR-CNN association head · KITTI MOTS val, Ped | sMOTSA: 43.2 -> 46.8 | Helped |
Closest papers by meaning
- Learning Multi-Object Tracking and Segmentation From Automatic Annotations (CVPR 2020 · similarity 0.94)
- DiffusionTrack: Diffusion Model for Multi-Object Tracking (AAAI 2024 · similarity 0.90)
- Multiple Object Tracking as ID Prediction (CVPR 2025 · similarity 0.89)
- OVT-B: A New Large-Scale Benchmark for Open-Vocabulary Multi-Object Tracking (NeurIPS 2024 · similarity 0.89)
- More Than Meets the Eye: Enhancing Multi-Object Tracking Even with Prolonged Occlusions (ICML 2025 · similarity 0.89)
- Focusing on Tracks for Online Multi-Object Tracking (CVPR 2025 · similarity 0.88)
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.