论文:http://arxiv.org/abs/1506.01497
1.准备工作
类似于fast-rcnn
http://blog.csdn.net/u010678153/article/details/46892911
2.安装(用于demo)
2.1 从github上clone到py-faster-rcnn的库。git clone --recursive https://github.com/rbgirshick/py-faster-rcnn.git
2.2建立模块
cd py-faster-rcnn/lib/ make
2.3 生成Caffe和pycaffe
cd $py-faster-rcnn/caffe-fast-rcnn
# Now follow the Caffe installation instructions here:
# http://caffe.berkeleyvision.org/installation.html
# If you're experienced with Caffe and have all of the requirements installed
# and your Makefile.config in place, then simply do:
make -j8 && make pycaffe
2.4 下载Faster RCNN检测器
cd $py-faster-rcnn
./data/scripts/fetch_faster_rcnn_models.sh
3.运行demo
如果报错可以看下是否支持vgg网络的显存以及demo里网络模型路径
cd py-faster-rcnn/
./tools/demo.py
4.训练
cd py-faster-rcnn/
./tools/train_faster_rcnn_alt_opt.py
一些可能根据自己的需要所修改的地方:./tools/train_faster_rcnn_alt_opt.py
parse_args()中weights的路径等参数,
最大迭代次数max_iters = [80000, 40000, 80000, 40000]
./lib/datasets
如果你的groundtruth和我一样并非xml文件,pascal_voc.py,imdb.py,factory.py修改成自己所需要的读入方式
如果有,可以照搬fast-rcnn的修改模式
./models/ZF/faster_rcnn_alt_opt (我用的zf模型)