Learning to Reconstruct 3D Manhattan Wireframes From a Single Image 环境配置与复现

conda create -y -n shapeunity-9.2　python=3.6
source activate shapeunity-9.2
# 我下面这句是一个一个下载的
conda install -y pyyaml docopt matplotlib scikit-image opencv tqdm
conda install -y pytorch==1.2.0 torchvision==0.4.0 cudatoolkit=9.2 -c pytorch
# 下面这句老失败，换成了下下句
python -m pip install --upgrade vispy cvxpy
pip install --upgrade vispy cvxpy -i https://pypi.tuna.tsinghua.edu.cn/simple

git clone https://github.com/zhou13/shapeunity
git clone https://github.com/moliushang/shapeunity
git clone git://github.com/zhou13/shapeunity
cd shapeunity
mkdir data logs

安装curl

source activate shapeunity-9.2

报错：

 curl --version
curl: symbol lookup error: /home/heixiazi/anaconda3/bin/../lib/libcurl.so.4: undefined symbol: SSLv2_client_method

解决：

# 定位
locate libcurl.so.4
sudo gedit ~/.bashrc

#　添加动态库
export LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH

# 让配置文件生效
source ~/.bashrc

运行：

python ./train.py -d 0,1,2,3 --identifier baseline config/hourglass.yaml

报错：ModuleNotFoundError: No module named 'yaml'

pip install pyyaml

报错：ModuleNotFoundError: No module named 'numpy'

pip install numpy -i https://pypi.tuna.tsinghua.edu.cn/simple

 报错：ModuleNotFoundError: No module named 'cv2'

conda uninstall opencv
pip install opencv-python -i https://pypi.tuna.tsinghua.edu.cn/simple 

报错：AttributeError: module 'yaml' has no attribute 'FullLoader'

1. 未安装 pyyaml

pip3 install pyyaml

2. 已存在pyyaml，且版本低于5.1

查看 yaml 版本，并卸载老版本：

    python3
    import yaml
    yaml.__version__
     

    # 这里用什么下载的就用什么卸载，比如我用conda下载的，这里我就必须用conda卸载
    conda uninstall pyyaml
    pip3 install pyyaml

报错：ModuleNotFoundError: No module named 'skimage'

pip install scikit-image　-i https://pypi.tuna.tsinghua.edu.cn/simple