Installation

Check INSTALL.md for installation instructions.

Perform training on Penn-Fudan for Pedestrian Detection dataset (https://www.cis.upenn.edu/~jshi/ped_html/) converted in COCO format for training. Dataset is placed in directory 'Human_detection_R2CNN/datasets/human'

1. start training. weight will save in directory 'R2CNN/output/human/' or we can change the path in config file 'Human_detection_R2CNN/configs/Human_det_train.yaml' in line no. 68

python tools/train_net.py --config-file "configs/Human_det_train.yaml"

Inference on Penn-Fudan for Pedestrian Detection dataset. Download trained model

1. Download model and put in drectory 'Human_detection_R2CNN/output/human/'

2. Single image inference. result will save 'Human_detection_R2CNN/datasets/human/result'

cd ./tools
python inference_engine.py