predict.py

import os
from cog import BasePredictor, Input, Path

import numpy as np
import torch
import torch.nn.functional as F
from torchvision.transforms.functional import normalize
from huggingface_hub import hf_hub_download
import gradio as gr
from briarmbg import BriaRMBG
import PIL
from PIL import Image
from typing import Tuple
    
def resize_image(image):
    image = image.convert('RGB')
    model_input_size = (1024, 1024)
    image = image.resize(model_input_size, Image.BILINEAR)
    return image

def inference(image, net):
    orig_image = Image.fromarray(image)
    w,h = orig_im_size = orig_image.size
    image = resize_image(orig_image)
    im_np = np.array(image)
    im_tensor = torch.tensor(im_np, dtype=torch.float32).permute(2,0,1)
    im_tensor = torch.unsqueeze(im_tensor,0)
    im_tensor = torch.divide(im_tensor,255.0)
    im_tensor = normalize(im_tensor,[0.5,0.5,0.5],[1.0,1.0,1.0])
    if torch.cuda.is_available():
        im_tensor=im_tensor.cuda()

    result=net(im_tensor)
    result = torch.squeeze(F.interpolate(result[0][0], size=(h,w), mode='bilinear') ,0)
    ma = torch.max(result)
    mi = torch.min(result)
    result = (result-mi)/(ma-mi)    
    im_array = (result*255).cpu().data.numpy().astype(np.uint8)
    pil_im = Image.fromarray(np.squeeze(im_array))
    new_im = Image.new("RGBA", pil_im.size, (0,0,0,0))
    new_im.paste(orig_image, mask=pil_im)
    return new_im

class Predictor(BasePredictor):
    def setup(self) -> None:
        self.net=BriaRMBG()
        model_path = hf_hub_download("briaai/RMBG-1.4", 'model.pth')
        if torch.cuda.is_available():
            self.net.load_state_dict(torch.load(model_path))
            self.net=self.net.cuda()
        else:
            self.net.load_state_dict(torch.load(model_path,map_location="cpu"))
        self.net.eval()
    def predict(
        self,
        input_image: Path = Input(description="Input Image"),
    ) -> Path:
        output_image = inference(input_image, self.net)
        return Path(output_image)