| import model | |
| from PIL import Image | |
| import torch | |
| device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') | |
| file = "./image.png" # input image | |
| model = model.BEN_Base().to(device).eval() #init pipeline | |
| model.loadcheckpoints("./BEN_Base.pth") | |
| image = Image.open(file) | |
| mask, foreground = model.inference(image) | |
| mask.save("./mask.png") | |
| foreground.save("./foreground.png") | |