#OpenCV_Unet This is the simple demo of the usage unet with opencv.

#Dependencies

• OpenCV 4.3.0 and higher preferable for CUDA backend usage.
• gcc/clang with c++17 support and higher.
• download models and train folder from https://mega.nz/folder/6i5llSqb#xqzNdy_jGYLTKNXtg94PBw and unpack to root of directory(files in the train folder will be appended)
• download video for example from https://www.youtube.com/c/FullRoadView/videos

#Tested

• Desktop Ubuntu 18.04
• Nvidia jetson nano
• MacOS

#Build Install opencv(for jetson nano there is a package which was prebuilt with CUDA support in the DNN module). Just run ./build_host.sh

#Run Run ./run.sh

Or you can use any your custom video/camera/video stream: ./build_host/opencv_unet <path to your video/camera device/videostream>