Download dlib (i.e. version 19.13) Link it to the opeye/dlib folder: ln -s /path/to/dlib-19.13/ dlib
Install opencv, then cuda & cudnn.
- mkdir build
- cd build
- cmake ..
Download & install cuda:
sudo dpkg -i cuda-repo-ubuntu1710-9-2-local_9.2.88-1_amd64.deb
sudo apt-key add /var/cuda-repo-<version>/7fa2af80.pub
sudo apt-get update
sudo apt-get install cuda
Make sure the right NVIDIA driver is installed (it will complain about unmet dependencies otherwise)
Download & install libcudnn runtime & development version (you need an nvidia account for this):
(on ubuntu make sure to use the deb packages => easiest)
- Started using the Window Development Env image.
- But it might be enough to install the "Visual Studio Build Tools" on your computer.
- Install git (git-scm.com)
- Install cmake (use MSI version, chose the "ass cmake to the system path for all (or current) users option")
- git clone
- cd to the repo
- Load json lib via: cd lib && mv json json.old && git clone https://github.com/nlohmann/json.git json (make sure there is a src folder => maybe rename include to src)
- download dlib / extract. DLib will be build together with the project. But if you want to test DLib, build as described here. (To use CUDA, install it before building, it is a must to train the neural networks)
- Download the lastest opencv source release. Extract to c:\opencv Build it (mkdir build && cd build && cmake .. && cmake --build . --config Release)
- Link the latest dlib version in th src folder. (Or extract it there)
- Download / install freeglut: (https://www.transmissionzero.co.uk/software/freeglut-devel/) extract it to c:
- Download / Install qt (use the online installer, install fhe latest prebuild for msvc 2017.
- mkdir build
- cd build
- cmake ../src/ -DCMAKE_PREFIX_PATH="C;\opencv\build;C:\freeglut;C:\Qt\5.11.0;C:\Qt\5.11.0\msvc2017_x64 -A x64"
- cmake --build . --config Release