app.py

import torch
from transformers import AutoModel, AutoTokenizer
from PIL import Image
import gradio as gr
import tempfile

# Load the OCR model and tokenizer
tokenizer = AutoTokenizer.from_pretrained('ucaslcl/GOT-OCR2_0', trust_remote_code=True)
model = AutoModel.from_pretrained('ucaslcl/GOT-OCR2_0', trust_remote_code=True, 
                                  low_cpu_mem_usage=True, 
                                  pad_token_id=tokenizer.eos_token_id).eval()

# Check if GPU is available and use it, else use CPU
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
model = model.to(device)

# Function to perform OCR on the image
def perform_ocr(image):
    # Save the image to a temporary file
    with tempfile.NamedTemporaryFile(suffix=".jpg", delete=False) as temp_file:
        image.save(temp_file.name)  # Save the image
        temp_image_path = temp_file.name  # Get the file path for the saved image

    # Perform OCR using the model
    result = model.chat(tokenizer, temp_image_path, ocr_type='ocr')
    return result

# Create the Gradio interface using the new syntax
interface = gr.Interface(
    fn=perform_ocr, 
    inputs=gr.Image(type="pil"),  # Updated to gr.Image
    outputs=gr.Textbox(),         # Updated to gr.Textbox
    title="OCR Web App",
    description="Upload an image to extract text using the GOT-OCR2.0 model."
)

# Launch the app
interface.launch()