The AI part is a tool called YOLO which runs primarily in Linux but there is a distribution for Windows.
You can find all the details here: https://github.com/unsky/yolo-for-windows-v2

Too hard to understand the instructions on the website?
IMPORTANT NOTE: Follow my instructions at your own risk. These are the links I used and I found no issue with them but please ensure your own protection against virus, dead links, weird links, etc.

1. Go to https://github.com/unsky/yolo-for-windows-v2 and click on clone or download green button, save as zip then unzip that in your desktop
1.1 Also download yolo.cfg and yolo-voc.cfg (a bit further down in the same page)
2. Download and install specifically CUDA 8.0 ( https://developer.nvidia.com/cuda-80-ga2-download-archive )
3. Download and install specifically MS Visual Studio 2015 ( https://go.microsoft.com/fwlink/?LinkId=532606&clcid=0x409 )
4. Download and unzip to the c: drive specifically OpenCV 2.4.9 https://sourceforge.net/projects/opencvlibrary/files/opencv-win/2.4.9/opencv-2.4.9.exe/download
5. Run Visual Studio (you need a microsoft account) and open the YOLO zip project you previously unziped in your desktop. This is the file you want to open: builddarknetdarknet.sln
6. On the menu bar you will find an option that says debug, change that to Release, and the one next to it that says Win32, change it to x64
7. Go to menu Build and press Build solution.
8. Now you have a new EXE file in your desktop YOLO folder in builddarknetx64Darknet.exe
9. Move these files ( opencv_core249.dll, opencv_highgui249.dll and opencv_ffmpeg249_64.dll ) from your OpenCV folder in C drive to your desktop YOLO folder in builddarknetx64
10. Also move yolo.weights and yolo-voc.weights to your desktop YOLO folder in builddarknetx64
11. You are so close! Now connect a webcam and run the darknet_web_cam_voc.cmd file in builddarknetx64
12. Blamo! You should be seeing some AI. If you have different webcams edit the darknet_web_cam_voc.cmd file and change the end number to the number webcam you want. Eg: darknet.exe detector demo data/voc.data yolo-voc.cfg yolo-voc.weights -c 1