intial commit

app.py

@@ -0,0 +1,72 @@
+import streamlit as st
+import numpy as np
+from PIL import Image
+from tensorflow.keras.models import load_model
+
+
+# Install the streamlit_drawable_canvas package if you haven't already
+# !pip install streamlit_drawable_canvas
+
+# Import the st_canvas function
+from streamlit_drawable_canvas import st_canvas
+
+# Function to preprocess the drawn image
+def preprocess_image(drawing, size=(28, 28)):
+    # Convert the drawing to a PIL Image
+    img = Image.fromarray(np.uint8(drawing))
+    # Resize the image to the desired size
+    img = img.resize(size)
+    # Convert the image to grayscale
+    img = img.convert('L')
+    # Convert the image to a numpy array
+    img_array = np.array(img)
+    # Normalize the pixel values to be between 0 and 1
+    img_array = img_array / 255.0
+    # Add a channel dimension (1 channel for grayscale)
+    img_array = np.expand_dims(img_array, axis=-1)
+    return img_array
+
+def preprocess_and_predict(img):
+    model = load_model("mnist_cnn_model.h5")
+    image = np.array(img) / 255.0
+    # Expand dimensions to match the input shape expected by the model
+    image = np.expand_dims(image, axis=0)
+    # Reshape to match the input shape expected by the model
+    image = np.reshape(image, (1, 28, 28, 1))
+    prediction = model.predict(image)
+    predicted_class = np.argmax(prediction)
+
+    return predicted_class
+
+# Main code
+def main():
+    st.title('Draw Digit')
+
+    # Create a drawing canvas
+    drawing = st_canvas(
+        fill_color="rgb(0, 0, 0)",  # Background color of the canvas
+        stroke_width=10,  # Stroke width
+        stroke_color="rgb(255, 255, 255)",  # Stroke color
+        background_color="#000000",  # Background color of the canvas component
+        height=150,  # Height of the canvas
+        width=150,  # Width of the canvas
+        drawing_mode="freedraw",  # Drawing mode: "freedraw" or "transform"
+        key="canvas",
+    )
+
+    # Check if the user has drawn anything
+    if drawing is not None:
+        st.image(drawing.image_data)
+
+        # Preprocess the drawn image
+        processed_image = preprocess_image(drawing.image_data)
+        st.write("Processed Image Shape:", processed_image.shape)
+
+        # Save the processed image
+        np.save("processed_image.npy", processed_image)
+        print(processed_image.shape)
+        digit_class = preprocess_and_predict(processed_image)
+        st.success(digit_class)
+
+if __name__ == "__main__":
+    main()

cnn_model_train.py

@@ -0,0 +1,64 @@
+# Importing necessary libraries
+import tensorflow as tf
+from tensorflow.keras import layers, models, datasets
+import numpy as np
+
+# Load the MNIST dataset
+(train_images, train_labels), (test_images, test_labels) = datasets.mnist.load_data()
+
+# Normalize pixel values to be between 0 and 1
+train_images = train_images.reshape((60000, 28, 28, 1)).astype('float32') / 255
+test_images = test_images.reshape((10000, 28, 28, 1)).astype('float32') / 255
+
+# Convert labels to categorical one-hot encoding
+train_labels = tf.keras.utils.to_categorical(train_labels, 10)
+test_labels = tf.keras.utils.to_categorical(test_labels, 10)
+
+# Define the CNN model
+def create_cnn_model(input_shape, num_classes):
+    model = models.Sequential()
+
+    # Convolutional layers
+    model.add(layers.Conv2D(32, (3, 3), activation='relu', input_shape=input_shape))
+    model.add(layers.MaxPooling2D((2, 2)))
+    model.add(layers.Conv2D(64, (3, 3), activation='relu'))
+    model.add(layers.MaxPooling2D((2, 2)))
+    model.add(layers.Conv2D(64, (3, 3), activation='relu'))
+
+    # Flatten layer to transition from convolutional layers to fully connected layers
+    model.add(layers.Flatten())
+
+    # Dense (fully connected) layers
+    model.add(layers.Dense(64, activation='relu'))
+    model.add(layers.Dense(num_classes, activation='softmax'))  # Output layer with softmax activation for multiclass classification
+
+    return model
+
+# Define input shape and number of classes
+input_shape = (28, 28, 1)  # Input shape for MNIST images
+num_classes = 10  # Number of classes for digit classification (0-9)
+
+# Create an instance of the model
+model = create_cnn_model(input_shape, num_classes)
+
+# Print model summary
+model.summary()
+
+# Compile the model
+model.compile(optimizer='adam',
+              loss='categorical_crossentropy',
+              metrics=['accuracy'])
+
+# Train the model
+model.fit(train_images, train_labels, epochs=5, batch_size=64, validation_data=(test_images, test_labels))
+# Save the trained model to disk
+model.save("mnist_cnn_model.h5")
+print("Model saved to disk.")
+
+# Load the saved model
+loaded_model = models.load_model("mnist_cnn_model.h5")
+print("Model loaded from disk.")
+
+# Evaluate the loaded model
+test_loss, test_accuracy = loaded_model.evaluate(test_images, test_labels)
+print(f"Test Accuracy: {test_accuracy}")

mnist_cnn_model.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6bd33517798011fbd1717bd9d757cfd2be2b1f286ce9a886d2b6b6a38c9ae865
+size 1172032

requirements.txt

@@ -0,0 +1,5 @@
+tensorflow
+numpy
+streamlit
+streamlit-drawable-canvas
+pillow