Research map / Detection and segmentation / Tracking
Memory and correlation-based MOT
Track multiple objects using correlation learning, memory, and trajectory prediction.
52 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Tracking, keyword cluster “mot · association · object tracking”
State History Benchmarks Tried Baselines Checklist Preprints Who Export
Where it stands
Few recent reports The tracked benchmarks have at most one report from 2025–26 that could beat the old record, too few to say whether they are still moving.
11 papers in 2023–24 and 9 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 16.4% to 11.2%.
Open questions the data points to
- Papers report Baseline on MOT17 validation (MOTA) anywhere from 49.33 to 69.1: comparisons against it depend on whose number you take.
- Papers report baseline on MOT17 validation (IDF1) anywhere from 58.91 to 72.9: comparisons against it depend on whose number you take.
No longer reported by papers here: MOT16 test (MOTA, last in 2021); MOT17 test (MOTA, last in 2024).
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Deep Tracking: Seeing Beyond Seeing Using Recurrent Neural Networks (AAAI 2016 · 181 citations)
- 2017 · Online Multi-Target Tracking Using Recurrent Neural Networks (AAAI 2017 · 527 citations)
- 2019 · Multi-Person Articulated Tracking With Spatial and Temporal Embeddings (CVPR 2019 · 96 citations)
- 2020 · PnPNet: End-to-End Perception and Prediction With Tracking in the Loop (CVPR 2020 · 183 citations)
- 2021 · Quasi-Dense Similarity Learning for Multiple Object Tracking (CVPR 2021 · 459 citations)
- 2022 · MeMOT: Multi-Object Tracking With Memory (CVPR 2022 · 238 citations)
- 2023 · MOTRv2: Bootstrapping End-to-End Multi-Object Tracking by Pretrained Object Detectors (CVPR 2023 · 255 citations)
- 2024 · SMILEtrack: SiMIlarity LEarning for Occlusion-Aware Multiple Object Tracking (AAAI 2024 · 83 citations)
- 2025 · Dual-Path Temporal Decoder for End-to-End Multi-Object Tracking (NeurIPS 2025 · 1 citation)
- 2026 · SAM2MOT: A Novel Paradigm of Multi-Object Tracking by Segmentation (AAAI 2026 · 6 citations)
Earliest papers here
- Deep Tracking: Seeing Beyond Seeing Using Recurrent Neural Networks (AAAI 2016 · 181 citations)
- The Solution Path Algorithm for Identity-Aware Multi-Object Tracking (CVPR 2016 · 41 citations)
- Online Multi-Target Tracking Using Recurrent Neural Networks (AAAI 2017 · 527 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
MOT16 test · MOTA (higher is better)
6 papers, 2017–2021 · not reported since 2021
| Year | Best so far | Set by |
|---|---|---|
| 2017 | 44.1 | Multi-Object Tracking With Quadruplet Convolutional Neural Networks (CVPR 2017 · 265 citations) |
| 2020 | 46.1 | DASOT: A Unified Framework Integrating Data Association and Single Object Tracking for Online Multi-Object Tracking (AAAI 2020 · 35 citations) |
| 2021 | 76.6 | Multiple Object Tracking With Correlation Learning (CVPR 2021 · 198 citations) |
MOT17 test · MOTA (higher is better)
5 papers, 2020–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 48 | DASOT: A Unified Framework Integrating Data Association and Single Object Tracking for Online Multi-Object Tracking (AAAI 2020 · 35 citations) |
| 2021 | 76.5 | Multiple Object Tracking With Correlation Learning (CVPR 2021 · 198 citations) |
| 2024 | 81.1 | SMILEtrack: SiMIlarity LEarning for Occlusion-Aware Multiple Object Tracking (AAAI 2024 · 83 citations) |
DanceTrack test · HOTA (higher is better)
4 papers, 2023–2025 · one report since 2024, not better
| Year | Best so far | Set by |
|---|---|---|
| 2023 | 69.9 | MOTRv2: Bootstrapping End-to-End Multi-Object Tracking by Pretrained Object Detectors (CVPR 2023 · 255 citations) |
What has been tried
45 papers here report 474 measured changes (a component added, removed or swapped, compared with a baseline): helped 77%, no clear effect 4%, hurt 18%, mixed 2%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add SLM to ByteTrack with YOLOX | MOT17 val · MOTA | 74.1 -> 76.2 | Helped | SMILEtrack: SiMIlarity LEarning for Occlusion-Aware Multiple Object Tracking (AAAI 2024 · 83 citations) |
| Replace D2MP with Kalman Filter | DanceTrack validation · HOTA | 55.7 -> 46.8 | Hurt | DiffMOT: A Real-time Diffusion-based Multiple Object Tracker with Non-linear Prediction (CVPR 2024 · 73 citations) |
| add MASA training to off-the-shelf SAM features | BDD100K MOT · AssocA | 32.9 -> 48.5 | Helped | Matching Anything by Segmenting Anything (CVPR 2024 · 41 citations) |
| Use path consistency loss instead of training with only regularization losses | MOT17 train-set evaluation · IDF1 | 3.4 -> 68.9 | Helped | Self-Supervised Multi-Object Tracking with Path Consistency (CVPR 2024 · 15 citations) |
| use static-background-only images instead of full images | VKITTI, sequence 01 · ATE mean (m) | 0.11860 -> 0.05340 | Helped | CTO-SLAM: Contour Tracking for Object-Level Robust 4D SLAM (AAAI 2024 · 6 citations) |
| add Hand-to-Object Attention | HOIST test · AP | 52.7 -> 56.4 | Helped | HOIST-Former: Hand-held Objects Identification Segmentation and Tracking in the Wild (CVPR 2024 · 2 citations) |
| add MaskObs to vanilla SSM query propagation | DanceTrack test · HOTA | 63.5 -> 64.8; +1.3 HOTA | Helped | Samba: Synchronized Set-of-Sequences Modeling for Multiple Object Tracking (ICLR 2025 · 1 citation) |
| Add instance-aware contrastive learning to hybrid matching | nuScenes val · AMOTA | 42.6% -> 44.7%; 2.1% | Helped | SynCL: A Synergistic Training Strategy with Instance-Aware Contrastive Learning for End-to-End Multi-Camera 3D Tracking (NeurIPS 2025 · 0 citations) |
Baseline numbers that disagree
The same baseline on the same benchmark and metric, reported more than 5% apart by different papers here. Training budgets, data splits and reimplementations differ, so state which number you compare with.
| Baseline | Benchmark | Metric | Reported | Papers |
|---|---|---|---|---|
| Baseline | MOT17 validation | MOTA | 49.33 – 69.1 | 3 papers, e.g. CVPR 2022 and CVPR 2021 |
| baseline | MOT17 validation | IDF1 | 58.91 – 72.9 | 2 papers, e.g. CVPR 2022 and CVPR 2021 |
| CenterTrack | MOT17 test | MOTA | 61.5 – 67.3 | 2 papers, e.g. CVPR 2021 and CVPR 2021 |
| FairMOT | MOT16 | MOTA | 69.3 – 74.9 | 2 papers, e.g. CVPR 2022 and AAAI 2022 |
What reviewers will expect
Datasets and baselines named in the 9 papers of this approach from 2024–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- MOT17 · 4 of 9 papers
- COCO · 3 of 9 papers
- DanceTrack · 2 of 9 papers
- SportsMOT · 2 of 9 papers
Baselines
- ByteTrack · 5 of 9 papers
- OC-SORT · 4 of 9 papers
- TransTrack · 2 of 9 papers
- MOTR · 2 of 9 papers
- QDTrack · 2 of 9 papers
- GTR · 2 of 9 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 20 papers; from OpenAlex and the papers' affiliations).
- Chinese Academy of Sciences · 2 papers
Use it
Download BibTeX of 33 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Association cues for online MOT (60 papers)
- Unified transformer tracking (25 papers)
Elsewhere in the topic
- Transformer-based video object tracking (74 papers)
- Foundation-model and correlation-filter tracking (66 papers)
- Siamese network tracking (34 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).