anyirao / video_unscreen Goto Github PK

Hi, thanks for your great work!

I notice that in configs/bg.json, the config of the pre-trained binary segmenter is
"binseg": {
"model_path": "./weights/exp-schp-201908261155-lip.pth",
"input_long_side": 962,
"stride_ratio": 0.5,
"flip": true
},
where the pre-trained model with file name "exp-schp-201908261155-lip.pth" seems not provided in the provided Google drive link(https://drive.google.com/drive/folders/1IYcUaimgllu_PyE6jBHDgn8FfD4stdkW?usp=sharing).
I tried all the provided pre-trained models but none of them match the structure of the human segmenter used in bg.py. I wonder if the pre-trained model is missed and if possible, could you upload that pre-trained model?

Looking forward to your reply