Submission details of Tracktor++_yw

Name Tracktor++_yw
Paper Link N/A
Code Link N/A
MOTA 0.308635
MOTP 0.235734
IDs 8808
False Positives 79973
False Negatives 572653
Input N/A
Runtime 0.5 s
Environment 1 GPU (Titan X)
Abstract The problem of tracking multiple objects in a video se- quence poses several challenging tasks. For tracking-by- detection, these include object re-identification, motion pre- diction and dealing with occlusions. We present a tracker (without bells and whistles) that accomplishes tracking without specifically targeting any of these tasks, in partic- ular, we perform no training or optimization on tracking data. To this end, we exploit the bounding box regression of an object detector to predict the position of an object in the next frame, thereby converting a detector into a Tracktor. We demonstrate the potential of Tracktor and provide a new state-of-the-art on three multi-object tracking benchmarks by extending it with a straightforward re-identification and camera motion compensation. We then perform an analysis on the performance and failure cases of several state-of-the-art tracking methods in comparison to our Tracktor. Surprisingly, none of the dedicated tracking methods are considerably better in deal- ing with complex tracking scenarios, namely, small and occluded objects or missing detections. However, our ap- proach tackles most of the easy tracking scenarios. There- fore, we motivate our approach as a new tracking paradigm and point out promising future research directions. Over- all, Tracktor yields superior tracking performance than any current tracking method and our analysis exposes remain- ing and unsolved tracking challenges to inspire future re- search directions.

Detailed results

Per-sequence results

sequence name MOTA MOTP IDs False Positives False Negatives
cubberly-auditorium-2019-04-22_1 44.21 24.64 102 1388 5892
discovery-walk-2019-02-28_0 35.57 23.80 120 741 7720
discovery-walk-2019-02-28_1 39.38 23.86 158 1046 7578
food-trucks-2019-02-12_0 50.10 21.57 391 1679 29917
gates-ai-lab-2019-04-17_0 34.83 18.44 167 5055 11606
gates-basement-elevators-2019-01-17_0 53.39 21.40 104 872 4805
gates-foyer-2019-01-17_0 60.12 19.30 57 1593 2413
gates-to-clark-2019-02-28_0 49.31 19.19 35 437 821
hewlett-class-2019-01-23_0 53.06 19.12 180 1706 5875
hewlett-class-2019-01-23_1 62.49 17.16 32 1299 596
huang-2-2019-01-25_1 30.26 21.34 46 1242 3336
huang-intersection-2019-01-22_0 20.21 26.34 372 4897 32144
indoor-coupa-cafe-2019-02-06_0 31.71 22.51 385 4094 40320
lomita-serra-intersection-2019-01-30_0 22.49 23.92 146 2272 13544
meyer-green-2019-03-16_1 19.29 26.18 260 2981 15653
nvidia-aud-2019-01-25_0 33.82 25.65 356 4411 17158
nvidia-aud-2019-04-18_1 41.81 18.41 35 1202 3409
nvidia-aud-2019-04-18_2 54.93 25.43 55 815 3719
outdoor-coupa-cafe-2019-02-06_0 15.59 26.24 517 8107 33716
quarry-road-2019-02-28_0 7.99 28.02 40 1966 3430
serra-street-2019-01-30_0 14.28 26.63 746 4417 31711
stlc-111-2019-04-19_1 68.10 18.54 43 490 1458
stlc-111-2019-04-19_2 57.41 18.97 36 617 893
tressider-2019-03-16_2 21.02 26.05 157 1695 19201
tressider-2019-04-26_0 24.81 25.40 1538 7934 80549
tressider-2019-04-26_1 30.30 22.73 1379 6359 126454
tressider-2019-04-26_3 30.78 26.21 1351 10658 68735
total 30.86 23.57 8808 79973 572653