Computer Vision and Touchdesigner
One of my main interests lately is the great power Computer Vision models have to enable real-time spatial interactions in physical venues. Since the release of Kinect-like cameras, artists and designers have used pose tracking models for interactivity.
But with the recent wave of AI models, hardware requirements to carry out these operations have practically disappeared.
Today, a simple home monitoring camera is enough to carry out complex scene understanding operations such as pose, face and gesture detection, or time-consistent video segmentation.