Creating an AI Director for Automating a Multi-Camera Setup | Mediapipe | OpenCV | (3/10)
This is a demo video of an application that processes streams from different cameras and detects facial pose, using mediapipe and some maths and then shifts the main OpenCV window based on which camera stream the user is looking at. This task of switching between live cameras is done by a Video Director or by a post-production editor, but here we’re doing with this with AI.
This is not just a python application, but it actually integrates with OBS studio, so you can directly use this in live Zoom, google hangouts, etc. calls or during a lecture recording.
The project is covered in detail in our upcoming course: Computer Vision For Building Cutting Edge Applications, you can join the course here: https://bleedaiacademy.com/computer-v...
Note: I was inspired by Google’s ML Solutions Architect Markku Lepistö project in which he did a similar thing but using a pose detection model, OBS python libs, and a google coral device.
#facelandmarksdetection #automation #mediapipe #opencv #python #machinelearning #deeplearning #computervision