ZHKKKe
commented
2 years ago Hi, thanks for your attention.
modnet_photographic_portrait_mattingis a model trained on 3,000 labeled portrait image data.modnet_webcam_portrait_mattingis a model that uses the proposed SOC strategy to fine-tune the modelmodnet_photographic_portrait_matting(on real-world unlabeled WebCam portrait data).
What's the difference between modnet_webcam_portrait_matting and modnet_photographic_portrait_matting? Are two different models used? Thanks!