caffe ssd

1.voc数据库下载

$ cd $HOME/data  
$ wget http://host.robots.ox.ac.uk/pascal/VOC/voc2012/VOCtrainval_11-May-2012.tar  
$ wget http://host.robots.ox.ac.uk/pascal/VOC/voc2007/VOCtrainval_06-Nov-2007.tar  
$ wget http://host.robots.ox.ac.uk/pascal/VOC/voc2007/VOCtest_06-Nov-2007.tar  
// Extract the data.  
$ tar -xvf VOCtrainval_11-May-2012.tar  
$ tar -xvf VOCtrainval_06-Nov-2007.tar  
$ tar -xvf VOCtest_06-Nov-2007.tar 

2.caffe-ssd 源码下载，编译

$ git clone https://github.com/weiliu89/caffe.git
$ sudo apt-get install python-opencv    //安装opencv2
//解决错误：ImportError: No module named caffe.proto
//在~/.bashrc文件末尾处添加如下内容：
// export PYTHONPATH=$PYTHONPATH:caffe目录/python
$ cp Makefile.config.example Makefile.config  
$ make all -j16   
$ make py  
$ make test -j16  
// (Optional)  
$ make runtest -j16 

3.net可视化

$ sudo -H pip install pydot      //需要安装pydot,否则画图会提示报错
$ sudo apt-get install graphviz  //否则会提示Exception: "dot" not found in path.
$ python python/draw_net.py xxx.prototxt xxx.png --rankdir=TB

4.创建lmdb文件

$ cd data/VOC0712
$ gvim create_list.sh
//修改: root_dir=$HOME/data/VOCdevkit/
//成：  root_dir=$HOME/data/VOCdevkit     然后保存
$ ./create_list.sh
$ ./create_data.sh
//会生成目录data/VOCdevkit/VOC0712

5.下载ssd预训练caffemodel

$ cp xxx/VGG_*.caffemodel model/VGG/
//到网站：http://download.csdn.net/download/zhayushui/10132277
//下载VGG_ILSVRC_16_layers_fc_reduced.caffemodel
//将VGG_*.caffemodel放到目录caffe/model/VGGNet

6.修改ssd 训练文件

$ cd caffe    //caffe 根目录
$ python example/ssd/ssd_pascal.py
//训练出来的*.caffemodel放在caffe/model/VGGNet/VOC0712

7.BBox-Label-Tool安装

//参考：https://www.cnblogs.com/objectDetect/p/5780006.html
$ mkdir BBox-Label-Tool
$ cd BBox-Label-Tool
$ git clone https://github.com/puzzledqs/BBox-Label-Tool .
$ python main.py
// sudo apt-get install python-tk     //解决ImportError:No Module named _tkinter问题