app.py

import streamlit as st
from PIL import Image, ImageOps, ImageEnhance
import requests
from transformers import TrOCRProcessor, VisionEncoderDecoderModel
import torch
import numpy as np

# Load pre-trained TrOCR model and processor
processor = TrOCRProcessor.from_pretrained("microsoft/trocr-base-handwritten")
model = VisionEncoderDecoderModel.from_pretrained("microsoft/trocr-base-handwritten")

# Function to perform OCR and extract text
def extract_text_from_image(image):
    # Preprocess the image
    image = preprocess_image(image)
    
    # Convert image to tensor
    pixel_values = processor(images=image, return_tensors="pt").pixel_values

    # Ensure the model is in evaluation mode
    model.eval()

    # Perform OCR
    with torch.no_grad():
        generated_ids = model.generate(pixel_values)
    
    # Decode the generated IDs to text
    text = processor.decode(generated_ids[0], skip_special_tokens=True)
    return text

# Image preprocessing function to enhance OCR performance
def preprocess_image(image):
    # Convert image to grayscale
    image = image.convert("RGB")
    image = ImageOps.grayscale(image)

    # Enhance the image for better contrast (optional)
    enhancer = ImageEnhance.Contrast(image)
    image = enhancer.enhance(2.0)  # Enhance contrast

    # Apply thresholding to make text more distinct (optional)
    image = image.point(lambda p: p > 200 and 255)  # Simple thresholding
    
    # Resize the image to a size appropriate for OCR
    image = image.resize((384, 384))  # Adjust this size as needed
    
    return image

# Streamlit UI
st.title("OCR Text Extraction from Image")

st.write("""
Upload an image containing text, and this app will extract and display the text from the image using the powerful TrOCR model!
""")

# File uploader to upload the image
uploaded_file = st.file_uploader("Choose an image...", type=["jpg", "jpeg", "png"])

if uploaded_file is not None:
    # Open the uploaded image
    image = Image.open(uploaded_file)

    # Display the uploaded image
    st.image(image, caption="Uploaded Image", use_column_width=True)

    # Button to extract text
    if st.button("Extract Text"):
        with st.spinner('Extracting text...'):
            extracted_text = extract_text_from_image(image)
            st.subheader("Extracted Text:")
            st.write(extracted_text)