I, too, have had fun playing with TensorFlow and MediaPipe to manipulate live video feed and change the background image. My experience is that MediaPipe's Self

I, too, have had fun playing with TensorFlow and MediaPipe to manipulate live video feed and change the background image. My experience is that MediaPipe's SelfieSegmentation model[1] is a lot quicker at identifying the foreground face than TensorFlow's body-pix[2] equivalent. Another approach I've tried is to use some form of chroma-key filter to identify the background (or foreground) in a video and then use the results of that (per frame) to either change the final output visually, or create clickable links from it[3]. All code in links below is Javascript: [1] - MediaPipe - https://codepen.io/kaliedarik/pen/PopBxBM https://codepen.io/kaliedarik/pen/PopBxBM [2] - TensorFlow - https://codepen.io/kaliedarik/pen/ZEeoZaP https://codepen.io/kaliedarik/pen/ZEeoZaP [3] - Chroma-key filter - https://scrawl-v8.rikweb.org.uk/demo/canvas-027.html https://scrawl-v8.rikweb.org.uk/demo/canvas-027.html