app.py

import gradio as gr
import torch
import pytesseract
import cv2
import tempfile
from transformers import AutoTokenizer, AutoModel

pytesseract.pytesseract.tesseract_cmd = r'/opt/homebrew/bin/tesseract'  # Update this if necessary

# Load the tokenizer and model
tokenizer_eng = AutoTokenizer.from_pretrained('ucaslcl/GOT-OCR2_0', trust_remote_code=True)
model_eng = AutoModel.from_pretrained('ucaslcl/GOT-OCR2_0', trust_remote_code=True).eval()

def perform_ocr(image, language):
    # Save the NumPy array as an image file temporarily
    with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as temp_file:
        temp_filename = temp_file.name
        cv2.imwrite(temp_filename, image)

    # Perform OCR for English
    res_eng = model_eng.chat(tokenizer_eng, temp_filename, ocr_type='ocr')

    # Clean up temporary file if needed
    # os.remove(temp_filename)

    return res_eng  # Return results for English

def ocr_and_search(image, language):
    # Call the perform_ocr function
    english_text = perform_ocr(image, language)
    
    return english_text  # Return the OCR result for English

# Create Gradio interface
iface = gr.Interface(
    fn=ocr_and_search,
    inputs=[
        gr.Image(type="numpy", label="Upload Image"),
        gr.Dropdown(choices=["English", "Hindi"], label="Select Language")
    ],
    outputs=gr.Textbox(label="Extracted Text"),
    title="OCR Application",
    description="Upload an image to extract text using OCR."
)

# Run the app
if __name__ == "__main__":
    iface.launch()