ECCV 2020 TAO Challenge Results

Click on a measure to sort the table accordingly. See below for a more detailed description.



TAO ECCV challenge deadline: August 16, 2020

Please submit a detailed description of your submission. If you would like to be considered for the “honorable mention” prize of $500, you must submit a short, 4-page manuscript detailing your approach to tao@motchallenge.net. We highly encourage all participants to do so!

Benchmark Statistics

TrackerTrackAP50TrackAP75TrackAP50_95TrackAP50_area_smallTrackAP50_area_mediumTrackAP50_area_largeHz
SORT_TAO
1. online method
10.217 4.383 4.892 7.720 8.227 15.166 16.3
A. Dave, T. Khurana, P. Tokmakov, C. Schmid, D. Ramanan. TAO: A Large-Scale Benchmark for Tracking Any Object. In The European Conference on Computer Vision (ECCV), 2020.
SequencesFramesTrajectoriesBoxes
122218467972164650

Difficulty Analysis

Sequence difficulty (from easiest to hardest, measured by average TrackAP50)

TAO_test

TAO_test

(10.217 TrackAP50)


Evaluation Measures

Lower is better. Higher is better.
Measure Better Perfect Description
TrackAP50 higher 1Track AP, IoU 0.5 [1]. Track AP at 3D IoU threshold of 0.5.
TrackAP75 higher 1Track AP, IoU 0.75 [1]. Track AP at 3D IoU threshold of 0.75.
TrackAP50_95 higher 1Track AP, IoU 0.5:0.95 [1]. Track AP averaged at 3D IoU thresholds of 0.5 to 0.95, with step size of 0.05.
TrackAP50_area_small higher 1Track AP, small objects [1]. Track AP at IoU threshold of 0.5 for objects with average area less than 32x32 pixels.
TrackAP50_area_medium higher 1Track AP, medium objects [1]. Track AP at IoU threshold of 0.5 for objects with average area between 32x32 and 96x96 pixels.
TrackAP50_area_large higher 1Track AP, large objects [1]. Track AP at IoU threshold of 0.5 for objects with average area over 96x96 pixels.
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
online method 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.
using public detections This method used the provided detection set as input.
using private detections This method used a private detection set as input.
new This entry has been submitted or updated less than a week ago.

References:


[1] Dave, A., Khurana, T., Tokmakov, P., Schmid, C. & Ramanan, D. TAO: A Large-Scale Benchmark for Tracking Any Object. In European Conference on Computer Vision, 2020.