Persolo_V2: Yolov4-based Pedestrian Detection

MOT20-06


Benchmark:

Short name:

Persolo_V2

Description:

Yolov4-based Pedestrian Detection trained on 200k images hydrated from OpenImage and fine-tuned on MOT20 Training Set.

Reference:

M. Elbatel, H. Maher, M. Abouzeid, A. Bayoumi. Persolo: A Pedestrian Is a Person Through Thick and Thin. In , 2022.

Last submitted:

September 06, 2021 (2 years ago)

Published:

February 27, 2023 at 12:54:29 CET

Submissions:

2

Open source:

No

Hardware:

Tesla T4 GPU

Runtime:

29.4 Hz

Benchmark performance:

Sequence AP MODA MODP FAF TP FP FN Rcll Prcn F1
MOT20Det0.8838.379.141.4317,139185,54726,38592.363.175.0

Detailed performance:

Sequence AP MODA MODP FAF TP FP FN Rcll Prcn F1
MOT20-040.9062.680.737.0221,53977,0099,19096.074.283.7
MOT20-060.74-17.374.063.352,72063,76411,16982.545.358.5
MOT20-070.9065.680.77.715,2074,5231,09193.377.184.4
MOT20-080.73-38.674.849.927,67340,2514,93584.940.755.1

Raw data: