MOT20 Results
Click on a measure to sort the table accordingly. See below for a more detailed description.
| Tracker | MOTA | IDF1 | HOTA | MOTP | MT | ML | FP | FN | Rcll | Prcn | AssA | DetA | AssRe | AssPr | DetRe | DetPr | LocA | FAF | ID Sw. | Frag | Hz | |
| ApLift 1.  |
58.9 ±20.3 | 56.5 | 46.6 | 79.6 | 513 (41.3) | 264 (21.3) | 17,739 | 192,736 | 62.8 | 94.8 | 45.2 | 48.2 | 48.1 | 76.8 | 51.3 | 77.5 | 82.2 | 4.0 | 2,241 (0.0) | 2,112 (0.0) | 0.4 | |
| MOTer 2.  |
58.6 ±16.4 | 46.7 | 40.1 | 79.8 | 441 (35.5) | 232 (18.7) | 33,691 | 175,841 | 66.0 | 91.0 | 32.7 | 49.7 | 40.6 | 57.3 | 54.4 | 75.0 | 82.2 | 7.5 | 4,850 (0.0) | 9,207 (0.0) | 1.0 | |
| Y. Xu, Y. Ban, G. Delorme, C. Gan, D. Rus, X. Alameda-Pineda. TransCenter: Transformers with Dense Queries for Multiple-Object Tracking. In arXiv, 2021. | ||||||||||||||||||||||
| MPNTrack 3. |
57.6 ±19.7 | 59.1 | 46.8 | 79.0 | 474 (38.2) | 279 (22.5) | 16,953 | 201,384 | 61.1 | 94.9 | 47.3 | 46.6 | 52.7 | 70.0 | 49.5 | 76.9 | 81.6 | 3.8 | 1,210 (19.8) | 1,420 (23.2) | 6.5 | |
| G. Brasó, L. Leal-Taixé. Learning a Neural Solver for Multiple Object Tracking. In CVPR, 2020. | ||||||||||||||||||||||
| TransCtr 4. |
57.6 ±17.9 | 47.2 | 39.8 | 79.4 | 442 (35.6) | 224 (18.0) | 39,820 | 174,838 | 66.2 | 89.6 | 32.4 | 49.2 | 40.1 | 57.3 | 54.4 | 73.6 | 81.8 | 8.9 | 4,915 (0.0) | 9,031 (0.0) | 1.0 | |
| Y. Xu, Y. Ban, G. Delorme, C. Gan, D. Rus, X. Alameda-Pineda. TransCenter: Transformers with Dense Queries for Multiple-Object Tracking. In arXiv, 2021. | ||||||||||||||||||||||
| TrTEST 5. |
57.0 ±20.4 | 48.8 | 42.5 | 79.7 | 499 (40.2) | 243 (19.6) | 26,770 | 190,486 | 63.2 | 92.4 | 38.0 | 47.9 | 40.4 | 79.0 | 51.8 | 75.8 | 82.3 | 6.0 | 5,271 (83.4) | 4,877 (77.2) | 1.0 | |
| center_reid 6. |
56.6 ±18.3 | 65.0 | 51.2 | 77.0 | 668 (53.8) | 141 (11.4) | 98,181 | 121,688 | 76.5 | 80.1 | 51.8 | 51.0 | 57.0 | 72.3 | 61.6 | 64.7 | 79.7 | 21.9 | 4,643 (60.7) | 9,188 (120.1) | 7.0 | |
| ALBOD 7. |
56.5 ±20.8 | 51.1 | 43.5 | 79.4 | 506 (40.7) | 228 (18.4) | 36,864 | 184,582 | 64.3 | 90.0 | 39.7 | 48.0 | 42.6 | 76.5 | 52.8 | 73.9 | 82.0 | 8.2 | 3,727 (57.9) | 4,308 (67.0) | 1.2 | |
| Anonymous submission | ||||||||||||||||||||||
| LPC_MOT 8. |
56.3 ±18.9 | 62.5 | 49.0 | 79.7 | 424 (34.1) | 313 (25.2) | 11,726 | 213,056 | 58.8 | 96.3 | 52.4 | 45.8 | 54.7 | 81.3 | 48.1 | 78.7 | 82.3 | 2.6 | 1,562 (26.6) | 1,865 (31.7) | 0.7 | |
| P. Dai, R. Weng, W. Choi, C. Zhang, Z. He, W. Ding. Learning a Proposal Classifier for Multiple Object tracking. In CVPR (Accepted), 2021. | ||||||||||||||||||||||
| FGRNetIV 9. |
55.4 ±17.6 | 52.7 | 42.5 | 79.4 | 508 (40.9) | 221 (17.8) | 38,973 | 189,715 | 63.3 | 89.4 | 39.3 | 46.3 | 46.4 | 62.2 | 51.4 | 72.5 | 81.8 | 8.7 | 2,159 (34.1) | 3,839 (60.6) | 1.3 | |
| GNNMatch 10. |
54.5 ±20.0 | 49.0 | 40.2 | 79.4 | 407 (32.8) | 317 (25.5) | 9,522 | 223,611 | 56.8 | 96.9 | 37.0 | 43.8 | 44.0 | 58.7 | 45.9 | 78.4 | 82.0 | 2.1 | 2,038 (35.9) | 2,456 (43.3) | 0.1 | |
| GCNNMatch: Graph Convolutional Neural Networks for Multi-Object Tracking via Sinkhorn Normalization. ARXIV 2020 | ||||||||||||||||||||||
| Tracker | MOTA | IDF1 | HOTA | MOTP | MT | ML | FP | FN | Rcll | Prcn | AssA | DetA | AssRe | AssPr | DetRe | DetPr | LocA | FAF | ID Sw. | Frag | Hz | |
| MOT20_TBC 11. |
54.5 ±17.2 | 50.1 | 0.0 | 77.3 | 415 (33.4) | 245 (19.7) | 37,937 | 195,242 | 62.3 | 89.5 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 100.0 | 8.5 | 2,449 (39.3) | 2,580 (41.4) | 5.6 | |
| Tracking-by-Counting: Using Network Flows on Crowd Density Maps for Tracking Multiple Targets[J]. IEEE Transactions on Image Processing, 2020, 30: 1439-1452. | ||||||||||||||||||||||
| UnsupTrack 12. |
53.6 ±19.0 | 50.6 | 41.7 | 80.1 | 376 (30.3) | 311 (25.0) | 6,439 | 231,298 | 55.3 | 97.8 | 40.2 | 43.3 | 43.2 | 75.9 | 45.1 | 79.8 | 82.6 | 1.4 | 2,178 (39.4) | 4,335 (78.4) | 1.3 | |
| S. Karthik, A. Prabhu, V. Gandhi. Simple Unsupervised Multi-Object Tracking. In Arxiv, 2020. | ||||||||||||||||||||||
| Tracktor++v2 13. |
52.6 ±19.9 | 52.7 | 42.1 | 79.9 | 365 (29.4) | 331 (26.7) | 6,930 | 236,680 | 54.3 | 97.6 | 42.0 | 42.3 | 45.9 | 71.6 | 44.1 | 79.3 | 82.4 | 1.5 | 1,648 (30.4) | 4,374 (80.6) | 1.2 | |
| P. Bergmann, T. Meinhardt, L. Leal-Taixé. Tracking without bells and whistles. In ICCV, 2019. | ||||||||||||||||||||||
| HOMI_Tracker 14. |
51.2 ±22.6 | 43.0 | 37.3 | 79.6 | 423 (34.1) | 312 (25.1) | 16,094 | 232,259 | 55.1 | 94.7 | 32.9 | 42.5 | 34.8 | 78.9 | 45.0 | 77.2 | 82.2 | 3.6 | 3,937 (71.4) | 6,458 (117.2) | 7.5 | |
| SFS 15. |
50.8 ±17.4 | 41.1 | 32.7 | 74.9 | 341 (27.5) | 251 (20.2) | 50,139 | 200,932 | 61.2 | 86.3 | 25.1 | 43.1 | 34.5 | 47.4 | 48.0 | 67.8 | 78.2 | 11.2 | 3,503 (57.3) | 7,617 (124.5) | 0.1 | |
| MTSFS： Online Multi-Object Tracking Based on Salient Feature Selection in Crowded Scenes | ||||||||||||||||||||||
| ITM 16. |
50.6 ±22.3 | 48.6 | 39.6 | 78.6 | 397 (32.0) | 320 (25.8) | 19,495 | 233,940 | 54.8 | 93.6 | 38.1 | 41.3 | 43.0 | 67.5 | 44.1 | 75.3 | 81.2 | 4.4 | 2,431 (44.4) | 4,533 (82.7) | 1.8 | |
| Pivot Correlation Network with Individualized Tubelets for Efficient Multi-object Tracking | ||||||||||||||||||||||
| BBT 17. |
46.8 ±21.3 | 42.2 | 35.8 | 78.0 | 312 (25.1) | 289 (23.3) | 35,014 | 236,176 | 54.4 | 88.9 | 32.2 | 40.2 | 35.1 | 70.2 | 43.8 | 71.7 | 80.6 | 7.8 | 3,880 (71.4) | 7,207 (132.6) | 8.0 | |
| MTAP-D-20-01870 Tracking Subjects and Detecting Relationships in Crowded City Videos (under review) | ||||||||||||||||||||||
| GMPHD_Rd20 18. |
44.7 ±20.7 | 43.5 | 35.6 | 77.5 | 293 (23.6) | 274 (22.1) | 42,778 | 236,116 | 54.4 | 86.8 | 32.0 | 39.9 | 34.1 | 72.0 | 44.0 | 70.3 | 80.2 | 9.6 | 7,492 (137.8) | 11,153 (205.1) | 25.2 | |
| Surveily 19. |
44.6 ±24.3 | 42.5 | 36.0 | 76.1 | 393 (31.6) | 236 (19.0) | 71,208 | 211,064 | 59.2 | 81.1 | 32.0 | 41.2 | 36.3 | 62.4 | 47.5 | 65.1 | 78.8 | 15.9 | 4,334 (73.2) | 6,646 (112.2) | 29.8 | |
| SORT20 20. |
42.7 ±18.6 | 45.1 | 36.1 | 78.5 | 208 (16.7) | 326 (26.2) | 27,521 | 264,694 | 48.8 | 90.2 | 35.9 | 36.7 | 39.4 | 66.9 | 39.5 | 73.0 | 81.0 | 6.1 | 4,470 (91.5) | 17,798 (364.4) | 57.3 | |
| A. Bewley, Z. Ge, L. Ott, F. Ramos, B. Upcroft. Simple online and realtime tracking. In 2016 IEEE International Conference on Image Processing (ICIP), 2016. | ||||||||||||||||||||||
| Sequences | Frames | Trajectories | Boxes |
| 4 | 4479 | 1501 | 765465 |
Difficulty Analysis
Sequence difficulty (from easiest to hardest, measured by average MOTA)
MOT20-04
(71.5 MOTA)
MOT20-07
(56.1 MOTA)
MOT20-06
(32.1 MOTA)
MOT20-08
(22.5 MOTA)
Evaluation Measures
Lower is better. Higher is better.| Measure | Better | Perfect | Description |
| MOTA | higher | 100% | Multi-Object Tracking Accuracy (+/- denotes standard deviation across all sequences) [1]. This measure combines three error sources: false positives, missed targets and identity switches. |
| IDF1 | higher | 100% | ID F1 Score [2]. The ratio of correctly identified detections over the average number of ground-truth and computed detections. |
| HOTA | higher | 100% | Higher Order Tracking Accuracy [3]. Geometric mean of detection accuracy and association accuracy. Averaged across localization thresholds. |
| MOTP | higher | 100% | Multi-Object Tracking Precision (+/- denotes standard deviation across all sequences) [1]. The misalignment between the annotated and the predicted bounding boxes. |
| MT | higher | 100% | Mostly tracked targets. The ratio of ground-truth trajectories that are covered by a track hypothesis for at least 80% of their respective life span. |
| ML | lower | 0% | Mostly lost targets. The ratio of ground-truth trajectories that are covered by a track hypothesis for at most 20% of their respective life span. |
| FP | lower | 0 | The total number of false positives. |
| FN | lower | 0 | The total number of false negatives (missed targets). |
| Rcll | higher | 100% | Ratio of correct detections to total number of GT boxes. |
| Prcn | higher | 100% | Ratio of TP / (TP+FP). |
| AssA | higher | 100% | Association Accuracy [3]. Association Jaccard index averaged over all matching detections and then averaged over localization thresholds. |
| DetA | higher | 100% | Detection Accuracy [3]. Detection Jaccard index averaged over localization thresholds. |
| AssRe | higher | 100% | Association Recall [3]. TPA / (TPA + FNA) averaged over all matching detections and then averaged over localization thresholds. |
| AssPr | higher | 100% | Association Precision [3]. TPA / (TPA + FPA) averaged over all matching detections and then averaged over localization thresholds. |
| DetRe | higher | 100% | Detection Recall [3]. TP /(TP + FN) averaged over localization thresholds. |
| DetPr | higher | 100% | Detection Precision [3]. TP /(TP + FP) averaged over localization thresholds. |
| LocA | higher | 100% | Localization Accuracy [3]. Average localization similarity averaged over all matching detections and averaged over localization thresholds. |
| FAF | lower | 0 | The average number of false alarms per frame. |
| ID Sw. | lower | 0 | Number of Identity Switches (ID switch ratio = #ID switches / recall) [4]. Please note that we follow the stricter definition of identity switches as described in the reference |
| Frag | lower | 0 | The total number of times a trajectory is fragmented (i.e. interrupted during tracking). |
| Hz | higher | Inf. | Processing speed (in frames per second excluding the detector) on the benchmark. The frequency is provided by the authors and not officially evaluated by the MOTChallenge. |
Legend
| Symbol | Description |
|
This is an online (causal) method, i.e. the solution is immediately available with each incoming frame and cannot be changed at any later time. |
|
This method used the provided detection set as input. |
 |
This method used a private detection set as input. |
 |
This entry has been submitted or updated less than a week ago. |
References:
| [1] | Evaluating Multiple Object Tracking Performance: The CLEAR MOT Metrics. Image and Video Processing, 2008(1):1-10, 2008. |
| [2] | Performance Measures and a Data Set for Multi-Target, Multi-Camera Tracking. In ECCV workshop on Benchmarking Multi-Target Tracking, 2016. |
| [3] | HOTA: A Higher Order Metric for Evaluating Multi-Object Tracking. International Journal of Computer Vision, 2020. |
| [4] | Learning to associate: HybridBoosted multi-target tracker for crowded scene. In Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2009. |
