app.py

import gradio as gr
import tensorflow as tf
import numpy as np
from PIL import Image
import io
import os
import requests
import tempfile

# Function to download the model from Hugging Face
def download_model_from_hf(model_path, local_dir):
    """Download model files from Hugging Face"""
    # Create a temporary directory to store the model
    os.makedirs(local_dir, exist_ok=True)
    
    # Extract the repository and file path from the URL
    # Example URL: https://huggingface.co/nivashuggingface/digit-recognition/resolve/main/saved_model
    parts = model_path.split('/')
    repo_id = f"{parts[3]}/{parts[4]}"
    file_path = '/'.join(parts[6:])
    
    # Download the model files
    api_url = f"https://huggingface.co/api/models/{repo_id}/revision/main/files/{file_path}"
    response = requests.get(api_url)
    
    if response.status_code == 200:
        # Download the saved_model.pb file
        saved_model_pb_url = f"https://huggingface.co/{repo_id}/resolve/main/{file_path}/saved_model.pb"
        pb_response = requests.get(saved_model_pb_url)
        if pb_response.status_code == 200:
            with open(os.path.join(local_dir, "saved_model.pb"), "wb") as f:
                f.write(pb_response.content)
        
        # Download the variables directory
        variables_dir = os.path.join(local_dir, "variables")
        os.makedirs(variables_dir, exist_ok=True)
        
        # Download variables.data-00000-of-00001
        variables_url = f"https://huggingface.co/{repo_id}/resolve/main/{file_path}/variables/variables.data-00000-of-00001"
        var_response = requests.get(variables_url)
        if var_response.status_code == 200:
            with open(os.path.join(variables_dir, "variables.data-00000-of-00001"), "wb") as f:
                f.write(var_response.content)
        
        # Download variables.index
        index_url = f"https://huggingface.co/{repo_id}/resolve/main/{file_path}/variables/variables.index"
        index_response = requests.get(index_url)
        if index_response.status_code == 200:
            with open(os.path.join(variables_dir, "variables.index"), "wb") as f:
                f.write(index_response.content)
        
        return True
    else:
        print(f"Failed to download model: {response.status_code}")
        return False

# Create a temporary directory for the model
MODEL_PATH = "https://huggingface.co/nivashuggingface/digit-recognition/resolve/main/saved_model"
LOCAL_MODEL_DIR = os.path.join(tempfile.gettempdir(), "digit_recognition_model")

# Download the model if it doesn't exist locally
if not os.path.exists(os.path.join(LOCAL_MODEL_DIR, "saved_model.pb")):
    print("Downloading model from Hugging Face...")
    download_model_from_hf(MODEL_PATH, LOCAL_MODEL_DIR)

# Load the model from local directory
print(f"Loading model from {LOCAL_MODEL_DIR}")
model = tf.saved_model.load(LOCAL_MODEL_DIR)

def preprocess_image(img):
    """Preprocess the drawn image for prediction"""
    # Convert to grayscale and resize
    img = img.convert('L')
    img = img.resize((28, 28))
    # Convert to numpy array and normalize
    img_array = np.array(img)
    img_array = img_array.astype('float32') / 255.0
    # Add batch dimension
    img_array = np.expand_dims(img_array, axis=0)
    # Add channel dimension
    img_array = np.expand_dims(img_array, axis=-1)
    return img_array

def predict_digit(img):
    """Predict digit from drawn image"""
    try:
        # Preprocess the image
        processed_img = preprocess_image(img)
        # Make prediction
        predictions = model(processed_img)
        predicted_digit = tf.argmax(predictions, axis=1).numpy()[0]
        # Get confidence scores
        confidence_scores = tf.nn.softmax(predictions[0]).numpy()
        # Create result string
        result = f"Predicted Digit: {predicted_digit}\n\nConfidence Scores:\n"
        for i, score in enumerate(confidence_scores):
            result += f"Digit {i}: {score:.2%}\n"
        return result
    except Exception as e:
        return f"Error during prediction: {str(e)}"

# Create Gradio interface
iface = gr.Interface(
    fn=predict_digit,
    inputs=gr.Image(type="pil", label="Draw a digit (0-9)"),
    outputs=gr.Textbox(label="Prediction Results"),
    title="Digit Recognition with CNN",
    description="""
    Draw a digit (0-9) in the box below. The model will predict which digit you drew.
    
    Instructions:
    1. Click and drag to draw a digit
    2. Make sure the digit is clear and centered
    3. The model will show the predicted digit and confidence scores
    """,
    examples=[
        ["examples/0.png"],
        ["examples/1.png"],
        ["examples/2.png"],
    ],
    theme=gr.themes.Soft(),
    allow_flagging="never"
)

# Launch the interface
if __name__ == "__main__":
    iface.launch()