app.py

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
from briarmbg import BriaRMBG
from PIL import Image
from fastapi import FastAPI, File, UploadFile
from fastapi.responses import FileResponse, JSONResponse
import os

app = FastAPI()

# 모델 로드
net = BriaRMBG()
model_path = hf_hub_download("briaai/RMBG-1.4", 'model.pth')
if torch.cuda.is_available():
    net.load_state_dict(torch.load(model_path, map_location="cuda", weights_only=True))
    net = net.cuda()
else:
    net.load_state_dict(torch.load(model_path, map_location="cpu", weights_only=True))
net.eval()

def resize_image(image):
    image = image.convert('RGB')
    model_input_size = (1024, 1024)
    image = image.resize(model_input_size, Image.BILINEAR)
    return image

def process_image(image: Image.Image):
    orig_image = image
    w, h = 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()

    # 모델 추론
    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)
    
    # 이미지 변환
    im_array = (result * 255).cpu().data.numpy().astype(np.uint8)
    pil_im = Image.fromarray(np.squeeze(im_array))
    new_im = Image.new("RGBA", orig_image.size, (0, 0, 0, 0))
    new_im.paste(orig_image, mask=pil_im)
    
    # 결과 이미지 저장
    output_path = "output_image.png"
    new_im.save(output_path)
    
    return output_path

@app.get("/")
def read_root():
    return {"message": "Welcome to the Background Removal API"}

@app.post("/remove-background/")
async def remove_background(file: UploadFile = File(...)):
    image = Image.open(file.file)
    output_path = process_image(image)
    return FileResponse(output_path, media_type="image/png", filename="output_image.png")

if __name__ == "__main__":
    import uvicorn
    uvicorn.run(app, host="0.0.0.0", port=7860)