Research map / Detection and segmentation
Tracking: research map
575 papers on Tracking in Detection and segmentation, from ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR (2016–2026), grouped into 2 clusters and 6 approaches. The busiest year so far is 2021.
What it is
Tracking follows objects through a video: one chosen object (single-object tracking) or every object of some classes (multi-object tracking, MOT). The core problems are finding the object again in each frame and linking detections over time despite occlusion.
Terms you will see on this page:
- Single-object tracking
- following one target given in the first frame.
- Multi-object tracking (MOT)
- detecting all objects and keeping their identities across frames.
- Data association
- deciding which detection in the new frame belongs to which track.
Start reading with the most cited papers here:
- High Performance Visual Tracking With Siamese Region Proposal Network (CVPR 2018 · 3,032 citations)
- Staple: Complementary Learners for Real-Time Tracking (CVPR 2016 · 1,851 citations)
Within Detection and segmentation, its share grew from 5.3% in 2023–24 to 6.6% in 2025–26 (67 → 80 papers at ICML, NeurIPS, ICLR, CVPR and AAAI, the venues with data for all four years).
Explore Tracking in the interactive map
Working on something in this topic? Describe your idea in scime atlas to see which approach it falls under, the closest papers by meaning and how crowded the spot has become.
What changed recently
Approaches whose share of the topic moved most between 2023–24 and 2025–26 (ICML, NeurIPS, ICLR, CVPR and AAAI):
- Unified transformer tracking rising: 4% → 11% of the topic (3 → 9 papers)
- Association cues for online MOT rising: 22% → 29% of the topic (15 → 23 papers)
- Transformer-based video object tracking rising: 36% → 40% of the topic (24 → 32 papers)
- Foundation-model and correlation-filter tracking falling: 15% → 8% of the topic (10 → 6 papers)
- Memory and correlation-based MOT falling: 16% → 11% of the topic (11 → 9 papers)
Papers from 2025–2026 that are already cited most:
- SUTrack: Towards Simple and Unified Single Object Tracking (AAAI 2025 · 50 citations)
- Two-stream Beats One-stream: Asymmetric Siamese Network for Efficient Visual Tracking (AAAI 2025 · 25 citations)
Approaches and key papers
Single-object tracking · 303 papers
Keywords: visual tracking, visual object, siamese
Approaches in this cluster:
- Foundation-model and correlation-filter tracking (123 papers)
Unify visual tracking with foundation models, diffusion, and overlap or correlation-filter trackers. - Siamese network tracking (70 papers)
Track objects by matching templates with Siamese networks, attention, and region proposals. - Transformer-based video object tracking (110 papers)
Tracks objects with video transformers, temporal context, point tracking and diffusion features.
Most cited and most cited since 2024:
- High Performance Visual Tracking With Siamese Region Proposal Network (CVPR 2018 · 3,032 citations)
- Staple: Complementary Learners for Real-Time Tracking (CVPR 2016 · 1,851 citations)
- OneTracker: Unifying Visual Object Tracking with Foundation Models and Efficient Tuning (CVPR 2024 · 139 citations)
- Event Stream-based Visual Object Tracking: A High-Resolution Benchmark Dataset and A Novel Baseline (CVPR 2024 · 69 citations)
Multi-object tracking · 272 papers
Keywords: mot, association, object tracking
Approaches in this cluster:
- Association cues for online MOT (107 papers)
Improve detection-to-track association with appearance, ID prediction, and weak cues. - Memory and correlation-based MOT (112 papers)
Track multiple objects using correlation learning, memory, and trajectory prediction. - Unified transformer tracking (53 papers)
Unify single and multi-object tracking with transformers and tracklet matching.
Most cited and most cited since 2024:
- Observation-Centric SORT: Rethinking SORT for Robust Multi-Object Tracking (CVPR 2023 · 1,003 citations)
- MOTS: Multi-Object Tracking and Segmentation (CVPR 2019 · 655 citations)
- Hybrid-SORT: Weak Cues Matter for Online Multi-Object Tracking (AAAI 2024 · 173 citations)
- SMILEtrack: SiMIlarity LEarning for Occlusion-Aware Multiple Object Tracking (AAAI 2024 · 83 citations)
Related topics in Detection and segmentation
- Open-set and few-shot detection (2,864)
- 3D detection and driving (761)
- CNN features (2,941)
- Weakly supervised detection and segmentation (791)
