app.py

from fastapi import FastAPI, File, UploadFile, HTTPException
from fastapi.responses import JSONResponse
from PIL import Image
import io
from omniparser import Omniparser
import os
import base64

# Initialize FastAPI app
app = FastAPI()

# Configuration for Omniparser
config = {
    'som_model_path': 'weights/icon_detect/best.pt',
    'device': 'cuda',  # Use 'cpu' if CUDA is unavailable
    'caption_model_path': 'weights/icon_caption_florence',
    'draw_bbox_config': {
        'text_scale': 0.8,
        'text_thickness': 2,
        'text_padding': 3,
        'thickness': 3,
    },
    'BOX_TRESHOLD': 0.05
}

# Initialize Omniparser
parser = Omniparser(config)


@app.post("/process/")
async def process_image(file: UploadFile = File(...)):
    """
    Endpoint to process an image and return parsed content.
    """
    try:
        # Load the uploaded image
        content = await file.read()
        image = Image.open(io.BytesIO(content))

        # Save image temporarily
        image_save_path = f"uploaded_images/{file.filename}"
        os.makedirs("uploaded_images", exist_ok=True)
        image.save(image_save_path)

        # Process the image using Omniparser
        image, parsed_content_list, label_coordinates = parser.parse(image_save_path)

        # # Convert processed image to base64
        # img_byte_arr = io.BytesIO()
        # image.save(img_byte_arr, format='PNG')
        # img_byte_arr = img_byte_arr.getvalue()
        # encoded_image = base64.b64encode(img_byte_arr).decode('utf-8')

        # Create response
        response = {
            "processed_image": image,
            "parsed_content": parsed_content_list,
            "label_coordinates": label_coordinates
        }

        return JSONResponse(content=response)

    except Exception as e:
        raise HTTPException(status_code=500, detail=f"Error processing image: {str(e)}")


@app.get("/")
async def root():
    return {"message": "OmniParser FastAPI is running"}