用Cornernet来检测自己的数据

Cornernet的安装 编译
1、Anaconda安装
wget …….......
bash Anaconda3-5.0.1-Linux-x86_64.sh 
source ~/.bashrc
2、Git clone https://github.com/princeton-vl/CornerNet.git
Cd CornerNet 
创建CornerNet环境
conda create --name CornerNet --file conda_packagelist.txt
激活
source activate CornerNet

cd <CornetNet dir>/models/py_utils/_cpools/
python setup.py install —user

cd <CornetNet dir>/external
make
会报错，缺少Cython，只需要conda install Cython

cd <CornetNet dir>/data
git clone https://github.com/cocodataset/cocoapi.git coco
cd <CornetNet dir>/data/coco/PythonAPI
make

下载 MS COCO Data
https://drive.google.com/file/d/1dop4188xo5lXDkGtOZUzy2SHOD_COXz4/view annotations.zip
并将annotations放置在<CornetNet dir>/data/coco
在<CornerNet dir>/data/coco/images/下新建三个文件夹trainval2014, minival2014 和 testdev2017将从http://cocodataset.org/#downlo