This repository contains source files of face detection using the FPN. It is developed based on the FPN network.
1、tensorflow
2、For training,you'd better use a GPU
git clone --recursive git@github.com:huangbo1221/FPN-face-detection.git
2、Download the pre-trained models named Resnet50.npy,password:kx45
3、Download the FPN-MODELS,password:e67q
After downloading the FPN-MODELS, place the weights file in FPN-face-detection/output/FPN_end2end/voc_2007_trainval. You should change some paths in FPN-face-detection/faster_rcnn/test_net.py. This file can use camera to capture your face,produce testing files in FDDB when you change inferences.
I use the wider-face dataset to train this network. Firstly you should transform this dataset to pascal voc dataset style.Tutorial can help us a lot.
ln -s your_Vocdevikit_path VOCdevkit2007
nohup ./experiments/scripts/FPN_end2end.sh 0 FPN pascal_voc2007 --set RNG_SEED 2 TRAIN.SCALES "[800]" > FPN.log 2>&1 &