app.py

import gradio as gr
import numpy as np
import pickle
from PIL import Image

# Load the model
with open('model.pkl', 'rb') as f:
    model_params = pickle.load(f)

W1 = model_params['W1']
b1 = model_params['b1']
W2 = model_params['W2']
b2 = model_params['b2']

def ReLu(Z):
    return np.maximum(Z, 0)

def softmax(Z):
    return np.exp(Z) / sum(np.exp(Z))

def forward_prop(W1, b1, W2, b2, X):
    Z1 = W1.dot(X) + b1
    A1 = ReLu(Z1)
    Z2 = W2.dot(A1) + b2
    A2 = softmax(Z2)
    return Z1, Z2, A1, A2

def get_predictions(A2):
    return np.argmax(A2, 0)

def preprocess_image(image):
    # Convert to grayscale
    img = image.convert('L')
    
    # Resize the image
    img = img.resize((28, 28))
    
    # Convert to numpy array and normalize
    img_array = np.array(img).reshape(1, 28*28) / 255.0
    
    return img_array.T  # Transpose to match the shape (784, 1)

def predict_digit(image):
    X = preprocess_image(image)
    
    # Forward propagation
    _, _, _, A2 = forward_prop(W1, b1, W2, b2, X)
    
    # Get the prediction
    prediction = get_predictions(A2)
    
    return int(prediction[0])

# Gradio interface
iface = gr.Interface(
    fn=predict_digit,
    inputs=gr.Image(type="pil"),
    outputs=gr.Label(num_top_classes=1),
    title="Handwritten Digit Recognition",
    description="Upload an image of a handwritten digit (0-9) and the model will predict which digit it is."
)

iface.launch()