Use VDO.Ninja with OpenCV, Tensorflow, and more
In this video we review the software and methods needed to bring WebRTC video (VDO.Ninja) into a Python script, as an accessible Numpy array. This means we can use low-latency streaming video as a raw image source for toolkits such as OpenCV, PyTorch, Tensorflow, and Scikit.
The latency is sub-second, and the code can run on an embedded linux systems (like a Raspberry Pi) or even Windows (via WSL). If you can run OpenCV-Python or Tensorflow already, then you likely have most the dependencies already installed.
0:00 Intro and demo
5:00 Running the code
7:50 Decoding WebRTC
14:30 Pushing into shared memory
22:30 Reading from shared memory
28:00 Running basic code example
30:15 Running advanced code example
34:00 Mjpeg streaming
40:30 Basic computer vision example
43:20 Core code snippet needed for your project
44:40 Another related project to check out
Find me on Discord @ https://discord.vdo.ninja
Raspberry.Ninja project @ https://raspberry.ninja/
Ooblex @ https://ooblex.com
Orange Pi 5 Plus 16GB @ https://www.aliexpress.com/item/10050...