faster-rcnn配置和訓練小結
阿新 • • 發佈:2018-12-29
論文:http://arxiv.org/abs/1506.01497
1.準備工作
類似於fast-rcnn
http://blog.csdn.net/u010678153/article/details/46892911
2.安裝(用於demo)
2.1 從github上clone到py-faster-rcnn的庫。git clone --recursive https://github.com/rbgirshick/py-faster-rcnn.git
2.2建立模組
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 下載Faster RCNN檢測器
cd $py-faster-rcnn
./data/scripts/fetch_faster_rcnn_models.sh
3.執行demo
如果報錯可以看下是否支援vgg網路的視訊記憶體以及demo裡網路模型路徑
cd py-faster-rcnn/
./tools/demo.py
4.訓練cd py-faster-rcnn/
./tools/train_faster_rcnn_alt_opt.py
一些可能根據自己的需要所修改的地方:./tools/train_faster_rcnn_alt_opt.py
parse_args()中weights的路徑等引數,
最大迭代次數max_iters = [80000, 40000, 80000, 40000]
./lib/datasets
如果你的groundtruth和我一樣並非xml檔案,pascal_voc.py,imdb.py,factory.py修改成自己所需要的讀入方式
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)