NOOUGAT: Towards Unified Online and Offline Multi-Object Tracking
Not on the map. Why?
Abstract (arXiv)
The long-standing division between textitonline and textitoffline Multi-Object Tracking (MOT) has led to fragmented solutions that fail to address the flexible temporal requirements of real-world deployment scenarios. Current textitonline trackers rely on frame-by-frame hand-crafted association strategies and struggle with long-term occlusions, whereas textitoffline approaches can cover larger time gaps, but still rely on heuristic stitching for arbitrarily long sequences. In this paper, we introduce NOOUGAT, the first tracker designed to operate with arbitrary temporal horizons. NOOUGAT leverages a unified Graph Neural Network (GNN) framework that processes non-overlapping subclips, and fuses them through a novel Autoregressive Long-term Tracking (ALT) layer. The subclip size controls the trade-off between latency and temporal context, enabling a wide range of deployment scenarios, from frame-by-frame to batch processing. NOOUGAT achieves state-of-the-art performance across both tracking regimes, improving textitonline AssA by +2.3 on DanceTrack, +9.2 on SportsMOT, and +5.0 on MOT20, with even greater gains in textitoffline mode.
Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Builds on 63 of its 101 references that are in the map. First posted on arXiv 2 Sep 2025.
Builds on
63 of 101 references.
References · 63
- 2014Adam: A Method for Stochastic Optimizationcited by 36,960
- 2020End-to-End Object Detection with Transformerscited by 7,384
- 2018BDD100K: A Diverse Driving Dataset for Heterogeneous Multitask Learningcited by 1,282
- 2021YOLOX: Exceeding YOLO Series in 2021cited by 1,064
- 2017Simple Online and Realtime Tracking with a Deep Association Metriccited by 908
- 2016Simple Online and Realtime Trackingcited by 862
- 2021ByteTrack: Multi-Object Tracking by Associating Every Detection Boxcited by 812
- 2016Performance Measures and a Data Set for Multi-Target, Multi-Camera Trackingcited by 685
- 2020HOTA: A Higher Order Metric for Evaluating Multi-Object Trackingcited by 546
- 2021TrackFormer: Multi-Object Tracking with Transformerscited by 515
- 2020FairMOT: On the Fairness of Detection and Re-Identification in Multiple Object Trackingcited by 489
- 2020Tracking Objects as Pointscited by 483
38 more
- 2021TransReID: Transformer-based Object Re-Identificationcited by 432
- 2019Tracking without bells and whistlescited by 382
- 2021MOTR: End-to-End Multiple-Object Tracking with Transformercited by 379
- 2020MOT20: A benchmark for multi object tracking in crowded scenescited by 355
- 2022Observation-Centric SORT: Rethinking SORT for Robust Multi-Object Trackingcited by 345
- 2022BoT-SORT: Robust Associations Multi-Pedestrian Trackingcited by 313
- 2018CrowdHuman: A Benchmark for Detecting Human in a Crowdcited by 283
- 2020Quasi-Dense Similarity Learning for Multiple Object Trackingcited by 220
- 2021DanceTrack: Multi-Object Tracking in Uniform Appearance and Diverse Motioncited by 213
- 2022StrongSORT: Make DeepSORT Great Againcited by 213
- 2019Learning a Neural Solver for Multiple Object Trackingcited by 198
- 2020FastReID: A Pytorch Toolbox for General Instance Re-identificationcited by 164
- 2022MOTRv2: Bootstrapping End-to-End Multi-Object Tracking by Pretrained Object Detectorscited by 148
- 2020MOTChallenge: A Benchmark for Single-Camera Multiple Target Trackingcited by 134
- 2023Deep OC-SORT: Multi-Pedestrian Tracking by Adaptive Re-Identificationcited by 130
- 2023SportsMOT: A Large Multi-Object Tracking Dataset in Multiple Sports Scenescited by 130
- 2020GNN3DMOT: Graph Neural Network for 3D Multi-Object Tracking with Multi-Feature Learningcited by 113
- 2022MeMOT: Multi-Object Tracking with Memorycited by 113
- 2022Global Tracking Transformerscited by 110
- 2020High-Order Information Matters: Learning Relation and Topology for Occluded Person Re-Identificationcited by 105
- 2020Unsupervised Pre-training for Person Re-identificationcited by 94
- 2018Features for Multi-Target Multi-Camera Tracking and Re-Identificationcited by 88
- 2023MeMOTR: Long-Term Memory-Augmented Transformer for Multi-Object Trackingcited by 86
- 2016Learning by tracking: Siamese CNN for robust target associationcited by 79
- 2017Tracking The Untrackable: Learning To Track Multiple Cues with Long-Term Dependenciescited by 76
- 2018Occluded Person Re-identificationcited by 72
- 2019How To Train Your Deep Multi-Object Trackercited by 71
- 2023Hybrid-SORT: Weak Cues Matter for Online Multi-Object Trackingcited by 66
- 2023MotionTrack: Learning Robust Short-term and Long-term Motions for Multi-Object Trackingcited by 64
- 2022Unifying Short and Long-Term Tracking with Graph Hierarchiescited by 62
- 2016Online Multi-Target Tracking Using Recurrent Neural Networkscited by 59
- 2024DiffMOT: A Real-time Diffusion-based Multiple Object Tracker with Non-linear Predictioncited by 59
- 2024Multiple Object Tracking as ID Predictioncited by 58
- 2021Exploring Simple 3D Multi-Object Tracking for Autonomous Drivingcited by 56
- 2020Lifted Disjoint Paths with Application in Multiple Object Trackingcited by 55
- 2021Learnable Graph Matching: Incorporating Graph Partitioning with Deep Feature Learning for Multiple Object Trackingcited by 47
- 2022Simple Cues Lead to a Strong Multi-Object Trackercited by 44
- 2021Learning a Proposal Classifier for Multiple Object Trackingcited by 42
50 most cited of 63 All 63 (JSON)
Built on by
Citing papers per year: 2025: 2; 2026: 1.
3 papers.