Win7 installation of Caffe and TensorFlow at the same time

　　

0: Today is 20171114, this tutorial does not know when to be effective

1: We installed the GPU version, so to install Cuda, this four platform compatible version is, cuda8+cudnn6. To the official website is the new version of CUDA8. Baidu Cloud: HTTP://PAN.BAIDU.COM/S/1PKKXS51,M2JL

2: Because Caffe out of Anaconda installation because of the lack of some dependency error patterns, we use the anaconda to configure the Python interface. Two are compatible, only the python3.5 version is supported. Baidu Cloud: Http://pan.baidu.com/s/1eSIo8gU,6jix

3: First installed Caffe, direct reference to the official website tutorial: https://github.com/BVLC/caffe/tree/windows. There are four points to note:

A:git checkout windows, after this command, the Caffe\scripts folder will be more than a build_win.cmd file, edit in the file, ... else (...). With_ninja=0......python_version=3.), no NINJA will compile slower but with possible errors

B:python interface Hereconda config --add channels conda-forge conda config --add channels willyd这两个最好换成 

Conda config--add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/conda config--add channels https ://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge Conda config--add channels https:// mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/msys2/this is replaced by the Tsinghua counterpart Mirror, the speed is relatively fast. In fact, many software Tsinghua have a corresponding image, you can try Google the corresponding keywords to find, ref:https://www.zhihu.com/question/38252144 C: Run Build_win.cmd will be downloaded Libraries_ V140_X64_PY35_1.1.0.TAR.BZ2 when the card master, at this time offline download----Baidu cloud: Http://pan.baidu.com/s/1eSIo8gU,6jix; put libraries_v140_x64_py35_ 1.1.0.tar.bz2 into C:\Users\Administrator\.caffe\dependencies\download, rerun Scripts\build_win.cmd D: There is a sentence in the document add the C:\Projects\caffe\pythonfolder to your Python path, "Your Python path" is adding pythonpath==....\caffe\ in the environment variable Python (the Python folder location you compiled in Caffe) 4: Install TensorFlow, reference https://www.tensorflow.org/install/install_windows;https:// zhuanlan.zhihu.com/p/29646950;http://www.jianshu.com/p/07130d272836, some online installation can not install the offline, Baidu Cloud: http://pan.baidu.com/ S/1pkkxs51,m2jl

Win7 installation of Caffe and TensorFlow at the same time