An openGL GLSL implementation of Dense Inverse Search Optical Flow with added STAF Openpose body tracking
We use vcpkg to install dependencies. Get vcpkg from the link and follow its installation instructions.
If you dont want to use vcpkg, then just ensure that the listed 3rd party libs are cmake findable on your system.
To make vcpkg use a little cleaner we set two environment variables, defining the tpe of system (x64 / x86) and the location of vcpkg.exe. Open a command promt with administrator privilages (hit windows key, type "cmd", right click "Command Prompt" and choose "Run as Administrator") . These commands may take a few seconds to execute.
setx VCPKG_DEFAULT_TRIPLET "x64-windows" /m
setx VCPKG_ROOT "C:\vcpkg" /m
Close the Admin Command Prompt window to flush the newly set variables.
Go to your vcpkg.exe installed location and open another command prompt.
vcpkg install glew glfw3 glm imgui realsense2 opencv
This should take 3-4 minutes.
To get timestamps from the realsense camera itself, rather than the time at which the host computer receives the frame we need to follow the steps described in these intel instructions under the section labeled "Enabling metadata on Windows"We use visual studio 2017 since it is the most readily available MSVC these days, support for c++17 features, and the hope that it will be useable with cuda 9.2.
We use cmake . Please use the latest version available.
Pull the latest version of oglFlowRealsense
git clone https://github.com/philipNoonan/oglFlowRS
Open CMake and set the source directory as "PATH_TO_YOUR_VERSION/oglFusionRealsense/" and the build directory as "PATH_TO_YOUR_VERSION/oglFusionRealsense/build"
Choose to create a new folder, and choose MSVC 15 2017 x64 as the generator.
Press "Configure"
Press "Generate"
Press "Open Project"
Easy bits:
sudo apt-get install libglew-dev
sudo apt-get install libglfw3 sudo apt-get install libglfw3-dev
sudo apt-get install libglm-dev
Less easy bits:
The sdk for the realsense can be downloaded and installed from source
OpenCV > 4.0 should also be installed from source
The STAF version of openpose is currently on this fork of the CMU openpose repo. The installation is the same as the openpose repo, but note that you should checkout the STAF branch.