在ubuntu下安装Detectron

安装Detectron

1）前提条件：

a) NVIDIA GPU, Linux, Python2
b) Caffe2, various standard Python packages, and the COCO API

2）检查caffe2是否安装

#交互Success，则为正常
python -c 'from caffe2.python import core' 2>/dev/null && echo "Success" || echo "Failure"
#返回可用GPU的个数，则为正常
python -c 'from caffe2.python import workspace; print(workspace.NumCudaDevices())'

若此步骤返回Failure或GPU个数小于1，则回到博客 中分析出错原因。

3）安装COCO API

# COCOAPI=你想把COCO克隆到本地的路径/cocoapi
git clone https://github.com/cocodataset/cocoapi.git $COCOAPI
cd $COCOAPI/PythonAPI
# Install into global site-packages
make install
# Alternatively, if you do not have permissions or prefer
# not to install the COCO API into global site-packages
python setup.py install --user

COCOAPI=你想把COCO克隆到本地的路径/cocoapi，在ubuntu终端中输入并赋值。

4）安装Detectron

a) 下载一个Detectron项目：

`# DETECTRON=/path/to/clone/detectron
git clone https://github.com/facebookresearch/detectron $DETECTRON

b) 安装你所下载的detectron目录中的requirements.txt

pip install -r $DETECTRON/requirements.txt

c) make

cd $DETECTRON && make

5) 查看是否Detectron安装成功

python $DETECTRON/detectron/tests/test_spatial_narrow_as_op.py

若交互以下内容，则表示detectron安装成功。