| MODNet webcam portrait matting model, safetensors format | |
| The model weights in model.safetensors are the official pre-trained MODNet webcam portrait matting | |
| checkpoint (modnet_webcam_portrait_matting.ckpt) by Zhanghan Ke, Jiayu Sun, Kaican Li, Qiong Yan and | |
| Rynson W.H. Lau, published at https://github.com/ZHKKKe/MODNet (commit | |
| 28165a451e4610c9d77cfdf925a94610bb2810fb). According to that repository's README, its code, models and demos | |
| are released under the Apache License 2.0 (see LICENSE). | |
| Modification: the checkpoint was converted from a PyTorch pickle to the safetensors format. All 751 | |
| state_dict entries keep their names and values. Tensors that the original checkpoint stores under two names | |
| (module.backbone.* and module.lr_branch.backbone.*) are saved as two identical copies. | |