OCSORTpublic: Observation-Centric SORT with public detections
Video not available.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Short name:
OCSORTpublic
Detector:
Public
Description:
A Kalman Filter- based multi-object tracker. We use only motion model without appearance features. The proposal of this method is to tackle multi-object tracking in occlusion and non-linear motion.
Reference:
J. Cao, X. Weng, R. Khirodkar, J. Pang, K. Kitani. Observation-Centric SORT: Rethinking SORT for Robust Multi-Object Tracking. In , 2022.
Last submitted:
March 29, 2022 (2 years ago)
Published:
March 29, 2022 at 03:24:45 CET
Submissions:
1
Project page / code:
n/a
Open source:
Yes
Hardware:
RTX 2080 i9@3GHZ
Runtime:
28.6 Hz
Benchmark performance:
| Sequence | MOTA | IDF1 | HOTA | MT | ML | FP | FN | Rcll | Prcn | AssA | DetA | AssRe | AssPr | DetRe | DetPr | LocA | FAF | ID Sw. | Frag |
| MOT17 | 58.2 | 65.1 | 52.4 | 432 (18.3) | 1,055 (44.8) | 4,379 | 230,449 | 59.2 | 98.7 | 57.6 | 47.8 | 63.5 | 76.4 | 49.5 | 82.6 | 84.3 | 0.2 | 784 (0.0) | 2,006 (0.0) |
Detailed performance:
| Sequence | MOTA | IDF1 | HOTA | MT | ML | FP | FN | Rcll | Prcn | AssA | DetA | AssRe | AssPr | DetRe | DetPr | LocA | FAF | ID Sw. | Frag |
| MOT17-01-DPM | 32.8 | 47.6 | 39.5 | 4 | 14 | 3 | 4,327 | 32.9 | 99.9 | 57.4 | 27.2 | 63.5 | 79.9 | 27.7 | 83.9 | 84.4 | 0.0 | 4 | 21 |
| MOT17-01-FRCNN | 40.2 | 52.1 | 42.9 | 5 | 13 | 8 | 3,843 | 40.4 | 99.7 | 55.4 | 33.2 | 60.8 | 80.4 | 34.0 | 83.8 | 84.6 | 0.0 | 5 | 32 |
| MOT17-01-SDP | 48.4 | 58.7 | 45.1 | 3 | 8 | 16 | 3,300 | 48.8 | 99.5 | 51.9 | 39.2 | 56.2 | 79.5 | 40.4 | 82.3 | 83.9 | 0.0 | 9 | 60 |
| MOT17-03-DPM | 73.6 | 78.9 | 61.6 | 76 | 19 | 891 | 26,661 | 74.5 | 98.9 | 63.5 | 59.9 | 68.3 | 80.3 | 62.3 | 82.7 | 84.0 | 0.6 | 34 | 177 |
| MOT17-03-FRCNN | 68.1 | 73.7 | 57.7 | 64 | 22 | 948 | 32,382 | 69.1 | 98.7 | 59.9 | 55.7 | 65.8 | 78.0 | 57.9 | 82.7 | 84.2 | 0.6 | 39 | 148 |
| MOT17-03-SDP | 83.5 | 85.6 | 67.2 | 103 | 15 | 1,325 | 15,905 | 84.8 | 98.5 | 67.7 | 66.9 | 72.4 | 81.0 | 70.3 | 81.7 | 83.6 | 0.9 | 56 | 193 |
| MOT17-06-DPM | 39.9 | 39.6 | 34.7 | 23 | 127 | 154 | 6,877 | 41.6 | 97.0 | 36.2 | 33.5 | 55.7 | 50.6 | 34.6 | 80.7 | 83.8 | 0.1 | 46 | 91 |
| MOT17-06-FRCNN | 45.8 | 47.0 | 38.5 | 29 | 105 | 163 | 6,168 | 47.7 | 97.2 | 39.1 | 38.0 | 56.2 | 56.3 | 39.5 | 80.6 | 83.7 | 0.1 | 61 | 109 |
| MOT17-06-SDP | 51.6 | 47.5 | 39.4 | 41 | 90 | 226 | 5,385 | 54.3 | 96.6 | 36.8 | 42.6 | 55.2 | 52.3 | 44.6 | 79.4 | 83.2 | 0.2 | 92 | 153 |
| MOT17-07-DPM | 37.6 | 37.5 | 31.2 | 1 | 29 | 23 | 10,475 | 38.0 | 99.6 | 31.3 | 31.1 | 36.7 | 59.8 | 31.9 | 83.6 | 85.1 | 0.0 | 44 | 92 |
| MOT17-07-FRCNN | 37.1 | 40.3 | 32.7 | 4 | 28 | 25 | 10,563 | 37.5 | 99.6 | 34.4 | 31.1 | 38.0 | 69.4 | 31.8 | 84.6 | 85.5 | 0.1 | 35 | 72 |
| MOT17-07-SDP | 47.9 | 46.5 | 37.2 | 9 | 17 | 140 | 8,599 | 49.1 | 98.3 | 34.7 | 40.1 | 38.4 | 68.2 | 41.5 | 83.1 | 85.0 | 0.3 | 64 | 126 |
| MOT17-08-DPM | 36.6 | 35.4 | 33.2 | 8 | 29 | 95 | 13,237 | 37.3 | 98.8 | 36.2 | 30.9 | 46.0 | 57.7 | 31.7 | 84.0 | 85.6 | 0.2 | 64 | 170 |
| MOT17-08-FRCNN | 32.2 | 33.9 | 32.5 | 7 | 37 | 77 | 14,202 | 32.8 | 98.9 | 39.0 | 27.4 | 48.7 | 60.4 | 28.0 | 84.6 | 85.9 | 0.1 | 37 | 111 |
| MOT17-08-SDP | 37.0 | 38.1 | 35.3 | 13 | 33 | 89 | 13,153 | 37.7 | 98.9 | 40.3 | 31.3 | 50.1 | 62.9 | 32.1 | 84.2 | 85.6 | 0.1 | 57 | 158 |
| MOT17-12-DPM | 45.8 | 58.2 | 47.6 | 10 | 50 | 74 | 4,608 | 46.8 | 98.2 | 57.1 | 39.6 | 62.1 | 79.8 | 40.9 | 85.8 | 87.4 | 0.1 | 14 | 35 |
| MOT17-12-FRCNN | 44.9 | 58.3 | 48.3 | 10 | 53 | 12 | 4,759 | 45.1 | 99.7 | 60.1 | 38.8 | 64.8 | 83.0 | 39.7 | 87.8 | 87.9 | 0.0 | 3 | 20 |
| MOT17-12-SDP | 49.9 | 60.1 | 50.8 | 13 | 49 | 22 | 4,311 | 50.3 | 99.5 | 60.5 | 42.8 | 66.4 | 78.3 | 44.0 | 87.1 | 87.5 | 0.0 | 5 | 27 |
| MOT17-14-DPM | 18.7 | 27.1 | 22.6 | 2 | 114 | 11 | 14,996 | 18.9 | 99.7 | 33.2 | 15.5 | 35.6 | 75.1 | 15.7 | 83.0 | 84.2 | 0.0 | 28 | 48 |
| MOT17-14-FRCNN | 20.4 | 27.7 | 23.0 | 3 | 118 | 15 | 14,661 | 20.7 | 99.6 | 31.4 | 16.8 | 34.7 | 72.1 | 17.1 | 82.3 | 84.2 | 0.0 | 33 | 52 |
| MOT17-14-SDP | 34.2 | 45.4 | 33.1 | 4 | 85 | 62 | 12,037 | 34.9 | 99.0 | 39.0 | 28.2 | 41.8 | 76.6 | 28.9 | 82.0 | 83.8 | 0.1 | 54 | 111 |
Raw data:
