I'm wanting to combine ffmpeg, scrcpy and face-api.node.js to achieve realtime facial recognition. I've currently got video running in the browser with openGL (demo https://giphy.com/gifs/U3uPdsKtCkaXdWYmiU ), but I'd like to remove the browser requirement (will not work in headless puppeteer) and do this in the Node backend. I'm able to grab screenshots and process those. I'm having trouble getting face-api backend to consume a scrcpy > ffmpeg stream -- not sure where the breakdown is. I'd like an output stream of face-api box coordinates, then send them back to WebsUp overlay with a websocket.
I'm wanting to combine ffmpeg, scrcpy and face-api.node.js to achieve realtime facial recognition. I've currently got video running in the browser with openGL (demo https://giphy.com/gifs/U3uPdsKtCkaXdWYmiU ), but I'd like to remove the browser requirement (will not work in headless puppeteer) and do this in the Node backend. I'm able to grab screenshots and process those. I'm having trouble getting face-api backend to consume a scrcpy > ffmpeg stream -- not sure where the breakdown is. I'd like an output stream of face-api box coordinates, then send them back to WebsUp overlay with a websocket.