Mask-Guided Attention Network for Occluded Pedestrian Detection

Pedestrian detection framework as detailed in arXiv report, accepted to ICCV 2019.

Installation

Our MGAN is based on mmdetection. Please check INSTALL.md for installation instructions.

The following commands will test the model on 1 GPU

./tools/dist_test.sh city_cfgs/mgan_50_65.py models/50_65.pth --eval bbox --out result/50_65.pkl

The following command will evaltate the results on CityPersons

python eval/eval_demo.py

Mask-Guided Attention Network for Occluded Pedestrian Detection. (ICCV'19)

Language:Python 90.7%Language:Cuda 6.4%Language:C++ 2.5%Language:Shell 0.4%