DcMOTv1: Jointing Recurrent Across-Channel and Spatial Attention for Multi-Object Tracking with Block-Erasing Data Augmentation


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.

Benchmark:

MOT16 | MOT17 | MOT20 |

Short name:

DcMOTv1

Detector:

Private

Description:

Reference:

Last submitted:

December 12, 2022 (3 months ago)

Published:

December 12, 2022 at 09:23:17 CET

Submissions:

2

Project page / code:

n/a

Open source:

No

Hardware:

GPU 1080TI

Runtime:

6.6 Hz

Benchmark performance:

Sequence MOTA IDF1 HOTA MT ML FP FN Rcll Prcn AssA DetA AssRe AssPr DetRe DetPr LocA FAF ID Sw. Frag
MOT2059.767.453.8828 (66.7)94 (7.6)106,62096,11081.479.855.053.060.572.465.664.380.723.85,636 (0.0)8,156 (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
MOT20-0479.581.063.75601832,08322,39591.888.762.964.768.876.674.872.282.015.41,8052,460
MOT20-0636.349.240.01224940,10042,17268.269.339.341.044.062.253.954.777.939.82,3293,438
MOT20-0773.667.155.37913,4214,93385.189.251.160.755.872.869.172.481.45.8386498
MOT20-0824.250.039.8672631,01626,61065.762.143.836.348.567.651.949.179.238.51,1161,760

Raw data: