ubuntu 16.04 + Anaconda2 + tensorflow-gpu 1.8 + caffe2 + Detectron

1. Anaconda2 安装

参见：Ubuntu18.04 安装 Anaconda3

2. tensorflow-gpu 1.8 安装

参见：Ubuntu18.04 + CUDA9.0 + cuDNN7.1.4 + TensorFlow1.8-python/anaconda 

3. caffe2 安装

参见：caffe2 Ubuntu16.04 + Anaconda 安装

4. Detectron安装

     a. 安装 COCO API

# COCOAPI=/path/to/clone/cocoapi 
# 我的是：COCOAPI=anaconda3/envs/anaconda2/lib/python2.7/site-packages/
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
python2 setup.py install --user

        b. 安装 Detectron

              1) 下载 Detectron repository

# DETECTRON=/path/to/clone/detectron
# 我的是：DETECTRON=anaconda3/envs/anaconda2/lib/python2.7/site-packages/
git clone https://github.com/facebookresearch/detectron $DETECTRON

              2）安装python依赖项

pip install -r $DETECTRON/requirements.txt

               3) 编译python模块

cd $DETECTRON && make

               4) 检验安装是否成功

python2 $DETECTRON/detectron/tests/test_spatial_narrow_as_op.py

5. 运行Detectron实例

首先，cd到detectron安装目录下，即tools文件夹的母目录，然后分别在终端输入以下指令，即可运行实例。

     a. Box 与 Mask

python2 tools/infer_simple.py \
    --cfg configs/12_2017_baselines/e2e_mask_rcnn_R-101-FPN_2x.yaml \ #model
    --output-dir /tmp/detectron-visualizations \ #输出文件夹
    --image-ext jpg \  #图像的文件扩展名
    --wts https://s3-us-west-2.amazonaws.com/detectron/35861858/12_2017_baselines/e2e_mask_rcnn_R-101-FPN_2x.yaml.02_32_51.SgT4y1cO/output/train/coco_2014_train:coco_2014_valminusminival/generalized_rcnn/model_final.pkl \  #weights
    demo #输入文件夹

      b. Box 与 Keypoints

python2 tools/infer_simple.py \
    --cfg configs/12_2017_baselines/e2e_keypoint_rcnn_R-101-FPN_s1x.yaml \
    --output-dir /tmp/detectron-visualizations \   #输出文件夹
    --image-ext jpg \
    --wts https://s3-us-west-2.amazonaws.com/detectron/37698009/12_2017_baselines/e2e_keypoint_rcnn_R-101-FPN_s1x.yaml.08_45_57.YkrJgP6O/output/train/keypoints_coco_2014_train%3Akeypoints_coco_2014_valminusminival/generalized_rcnn/model_final.pkl \
    demo   #输入文件夹