Clone the face Faster R-CNN repository
Git clone –recursive https://github.com/playerkk/face-py-faster-rcnn.git
Build the Cython modules
cd $FRCN_ROOT/lib
make
Build Caffe and pycaffe
cd $FRCN_ROOT/caffe-fast-rcnn
make -j8 && make pycaffe
下载预先训练好的VGG模型
A pre-trained face detection model trained on the WIDER training set is available here.
http://supermoe.cs.umass.edu/%7Ehzjiang/data/vgg16_faster_rcnn_iter_80000.caffemodel
放置目录:
$FRCN_ROOT/output/faster_rcnn_end2end/train/vgg16_faster_rcnn_iter_80000.caffemodel
下载测试数据
下载FDDB数据库放入$FRCN_ROOT/data目录:
包括:
FDDB
FDDB/FDDB-folds
FDDB/originalPics
python ./tools/run_face_detection_on_fddb.py --gpu=0
遇到的问题:
1:安装 Cython,python-opencv,easydict;不然会提示安装出错的;
可以直接下载easydict,然后tar -zxvf easydict.1.7.0.tar.gz; cd easydict.1.7.0; python setup.py install
解决即可正常的运行:
2:caffe设置的问题
1配置python layers
#In your Makefile.config, make sure to have this line uncommented
WITH_PYTHON_LAYER := 1
# Unrelatedly, it's also recommended that you use CUDNN
USE_CUDNN := 1 #可以注释掉的
Download pre-computed Faster R-CNN detectors
cd $FRCN_ROOT
./data/scripts/fetch_faster_rcnn_models.sh //也可官网上下载相应的faster_rcnn_models.tgz;然后自己解压;
Download the WIDER face dataset. Extract all files into one directory named WIDER
http://mmlab.ie.cuhk.edu.hk/projects/WIDERFace/
WIDER/
WIDER/WIDER_train/
WIDER/WIDER_val/
Download the (http://jianghz.me/files/wider_face_train_annot.txt) and put it under the WIDER directory.
Create symlinks for the WIDER dataset
cd $FRCN_ROOT ln -s $WIDER ~/WIDER //WIDER在home目录下
Follow the next sections to download pre-trained ImageNet models
cd $FRCN_ROOT
./data/scripts/fetch_imagenet_models.sh //也可官网上下载相应的imagenet_models.tgz;然后自己解压;
To train a Faster R-CNN face detector using the approximate joint training method, use experiments/scripts/faster_rcnn_end2end.sh. Output is written underneath $FRCN_ROOT/output.
cd FRCN_ROOT
./experiments/scripts/faster_rcnn_end2end.sh [GPU_ID] [NET] wider [–set …]
eg:
./experiments/scripts/faster_rcnn_end2end.sh 0 VGG16 wider
这是由于numpy的版本太高,numpy 1.12.0对这个做了些调整,把numpy降级到1.11.0就行了。
sudo pip install -U numpy==1.11.0
同时也可以下载numpy.1.11.0,自己安装一下;就可以跟新到1.11.0版本;
但是我的服务器中有两个python,2.7和3.4,而系统默认pip是装在python3.4上的,这样可以看见:
输入:pip --version
显示:pip 9.0.1 from /usr/local/lib/python3.4/dist-packages (python 3.4)
所以执行以下代码,装到强制装到python2.7中:
sudo python2.7 /usr/local/bin/pip install -U numpy==1.11.0
至此py-faster-rcnn在我这儿可以顺利训练了:
第二个问题:提示找不到图片的问题
主要是数据集的图片路径最后会多一个\r,这都是英文下载的图片标签windows下和Linux下多一个换行符的问题;
提取钱len-1个字符就可以了,可以在face.py中修改image_name这个字符串;
参考博客:http://blog.csdn.net/zengdong_1991/article/details/66475821