CLIPSeg Text-Prompted Segmentation

Choose a photo, capture one with your camera, drop or paste an image, or try the example, then type what to find in it ("a dog", "the sky"), and CLIPSeg masks it, with no fixed list of classes. It runs in your browser on LiteRT.js: all three graphs on WebGPU when it is available, on WebAssembly…

Description excerpted from the original listing, which is linked below.