app.py

import os
import streamlit as st
import requests
from PIL import Image
import numpy as np
from transformers import AutoProcessor, AutoModelForPreTraining
import torch
from huggingface_hub import login

# Get the token from environment variables (set in Hugging Face Space)
token = os.getenv("HF_Token")
login(token)

# Initialize the Hugging Face model
tokenizer = AutoProcessor.from_pretrained("google/paligemma-3b-mix-224")
model = AutoModelForPreTraining.from_pretrained("google/paligemma-3b-mix-224")

# Function to transcribe handwritten notes using Hugging Face model
def transcribe_handwriting(image):
    # Convert image to array and preprocess
    image = image.convert("RGB")
    image = np.array(image)

    # Prepare input for the model
    inputs = tokenizer(image, return_tensors="pt")
    
    # Generate output
    with torch.no_grad():
        outputs = model.generate(**inputs, max_length=512)
    
    transcription = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return transcription

# Set Streamlit page configuration
st.set_page_config(
    page_title="Prescription Reader",
    page_icon="💊",
    layout="centered",
)

# Header
st.title("Doctor's Prescription Reader 💊")

# Upload prescription image
uploaded_file = st.file_uploader("Upload Prescription Image", type=["jpg", "jpeg", "png"])

if uploaded_file is not None:
    # Display uploaded image
    st.image(uploaded_file, caption="Uploaded Prescription", use_column_width=True)

    with st.spinner("Transcribing handwriting..."):
        # Transcribe handwritten notes
        extracted_text = transcribe_handwriting(uploaded_file)
        st.subheader("Transcribed Text from Prescription:")
        st.text(extracted_text)