\n\n# Defining MNIST Data File Paths 📁\n\nIn this cell, we define the file paths for the MNIST dataset's training and testing images and labels using `os.path.join()` for compatibility across systems.\n\n- `train-images.idx3-ubyte` and `train-labels.idx1-ubyte` are for the training data.\n- `t10k-images.idx3-ubyte` and `t10k-labels.idx1-ubyte` are for the test data.\n\nThese paths will be used to load the dataset. 🔄","metadata":{"id":"ciozjwERJ176"}},{"cell_type":"code","source":"# Define the file paths for the training and testing data\ntrain_image = os.path.join(path, 'train-images.idx3-ubyte') # Path to the training images file\ntrain_label = os.path.join(path, 'train-labels.idx1-ubyte') # Path to the training labels file\n\ntest_image = os.path.join(path, 't10k-images.idx3-ubyte') # Path to the test images file\ntest_label = os.path.join(path, 't10k-labels.idx1-ubyte') # Path to the test labels file","metadata":{"id":"-8kEE337JZDT","trusted":true,"execution":{"iopub.status.busy":"2025-02-13T23:43:38.584319Z","iopub.execute_input":"2025-02-13T23:43:38.584705Z","iopub.status.idle":"2025-02-13T23:43:38.590350Z","shell.execute_reply.started":"2025-02-13T23:43:38.584664Z","shell.execute_reply":"2025-02-13T23:43:38.589057Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
\n
📂 MNIST Data Loader Class
\n
This class is responsible for loading and parsing the MNIST dataset from binary files.
\n
\n","metadata":{"id":"uoB_fyxzFsxj"}},{"cell_type":"code","source":"import struct # For handling binary data\nimport numpy as np # NumPy for array operations\nfrom array import array # Array module for handling byte data\n\nclass MnistDataLoader:\n \"\"\"\n A class to load the MNIST dataset from binary files.\n \"\"\"\n\n def __init__(self, train_image, train_label, test_image, test_label):\n self.train_image = train_image\n self.train_label = train_label\n self.test_image = test_image\n self.test_label = test_label\n\n def read_images_labels(self, img_path, label_path):\n \"\"\"Reads images and labels from the given file paths.\"\"\"\n\n # Read labels\n with open(label_path, 'rb') as file:\n magic, size = struct.unpack(\">II\", file.read(8))\n if magic != 2049:\n raise ValueError(f'Magic number mismatch, expected 2049, got {magic}')\n labels = np.array(array(\"B\", file.read()))\n\n # Read images\n with open(img_path, 'rb') as file:\n magic, size, rows, cols = struct.unpack(\">IIII\", file.read(16))\n if magic != 2051:\n raise ValueError(f'Magic number mismatch, expected 2051, got {magic}')\n image_data = np.array(array(\"B\", file.read())).reshape(size, rows, cols)\n\n return image_data, labels\n\n def load_data(self):\n \"\"\"Loads training and test data.\"\"\"\n x_train, y_train = self.read_images_labels(self.train_image, self.train_label)\n x_test, y_test = self.read_images_labels(self.test_image, self.test_label)\n return (x_train, y_train), (x_test, y_test)","metadata":{"id":"Bt61v0TgLTmA","trusted":true,"execution":{"iopub.status.busy":"2025-02-13T23:43:38.591693Z","iopub.execute_input":"2025-02-13T23:43:38.592096Z","iopub.status.idle":"2025-02-13T23:43:38.612318Z","shell.execute_reply.started":"2025-02-13T23:43:38.592061Z","shell.execute_reply":"2025-02-13T23:43:38.611087Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
\n\n# Simulating MNIST Dataset Loading 🚀\n\nThis code simulates the loading process of the MNIST dataset with a progress bar. It includes the following steps:\n\n1. **Progress Bar Simulation**: A simple loop simulates the loading process of the MNIST dataset. The progress is displayed in percentage, updating every second.\n\n2. **DataLoader Initialization**: Once the progress bar completes, the `MnistDataLoader` object is initialized with training and testing data (`train_image`, `train_label`, `test_image`, `test_label`). This object is assumed to handle data loading.\n\n3. **Loading Data**: The `Load_Data()` method of the `MnistDataLoader` class is called to fetch the training and test data, which are then unpacked into variables `x_train`, `y_train`, `x_test`, and `y_test`.\n\n4. **Completion Message**: After the data is successfully loaded, a confirmation message `MNIST loaded` is printed.\n\nThis approach helps in visualizing the progress during the dataset loading process while preparing the data for machine learning models.\n","metadata":{"id":"K1-fc1VeHHCE"}},{"cell_type":"code","source":"import time\n\n# Starting the process of loading the MNIST dataset\nprint('MNIST Loading ...')\n\n# Total number of steps for the loading simulation\ntotal_steps = 5\n\n# Loop to simulate the loading process with a progress bar\nfor i in range(total_steps):\n # Calculate the completion percentage\n percent = (i + 1) * 100 / total_steps\n\n # Print the current progress on the same line\n print(f\"\\r{percent:.1f}% Loading\", end=\"\", flush=True)\n\n # Simulate loading delay\n time.sleep(1)\n\n# Assuming MnistDataLoader is defined and takes the training and test data as input\nmnist_dataloader = MnistDataLoader(train_image, train_label, test_image, test_label)\n\n# Loading the data through the MnistDataLoader object\n(x_train, y_train), (x_test, y_test) = mnist_dataloader.load_data() # Changed Load_Data to load_data\n# Print a message indicating successful loading of the MNIST dataset\nprint(\"\\nMNIST loaded \")","metadata":{"id":"bhxvauGEr0Ik","outputId":"22bb1bf1-5552-42f4-965d-d278244b593f","trusted":true,"execution":{"iopub.status.busy":"2025-02-13T23:43:38.613491Z","iopub.execute_input":"2025-02-13T23:43:38.613863Z","iopub.status.idle":"2025-02-13T23:43:44.241429Z","shell.execute_reply.started":"2025-02-13T23:43:38.613829Z","shell.execute_reply":"2025-02-13T23:43:44.240222Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
\n\n# Data Normalization and Reshaping 🔄\n\nAfter loading the MNIST dataset, we perform two key preprocessing steps:\n\n1. **Normalization**: The pixel values of the images are scaled to the range [0, 1] by dividing by 255.0. This ensures that the values are in a suitable range for neural network training.\n \n2. **Reshaping**: The images are reshaped to have the shape `(28, 28, 1)` to match the input dimensions expected by Convolutional Neural Networks (CNNs), where:\n - `28 x 28` represents the image dimensions.\n - `1` indicates that the images are grayscale (one color channel).\n\nThese steps are essential to prepare the data for model training. 🧠⚙️","metadata":{"id":"Urpm0XyjHyu3"}},{"cell_type":"code","source":"# Normalize the pixel values to be in the range [0, 1]\nx_train, x_test = x_train / 255.0, x_test / 255.0\n\n# Reshape the data to match the input shape for CNNs (28x28 pixels, 1 channel)\nx_train = x_train.reshape(-1, 28, 28, 1)\nx_test = x_test.reshape(-1, 28, 28, 1)","metadata":{"id":"cSvoFdw4wNy6","trusted":true,"execution":{"iopub.status.busy":"2025-02-13T23:43:44.242685Z","iopub.execute_input":"2025-02-13T23:43:44.242980Z","iopub.status.idle":"2025-02-13T23:43:44.624257Z","shell.execute_reply.started":"2025-02-13T23:43:44.242955Z","shell.execute_reply":"2025-02-13T23:43:44.622965Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
\n\n# Building the Convolutional Neural Network (CNN) Model 🧠🔧\n\nIn this cell, we define the structure of our CNN model using Keras:\n\n1. **Convolutional Layers**:\n - The model starts with a series of convolutional layers (Conv2D) to extract features from the input images. Each convolutional layer is followed by a ReLU activation function.\n - The filters in the convolutional layers increase as we progress deeper into the model (32 → 64 → 64), which allows the model to capture more complex patterns.\n\n2. **Max Pooling Layers**:\n - After each convolutional layer, we use max pooling layers (MaxPooling2D) with a 2x2 pool size to reduce the spatial dimensions of the feature maps and prevent overfitting.\n\n3. **Fully Connected Layers**:\n - After the feature extraction, the output is flattened into a 1D vector and passed through a fully connected layer (Dense) with 64 neurons and ReLU activation.\n \n4. **Output Layer**:\n - The final layer has 10 neurons, corresponding to the 10 classes in the MNIST dataset, with a softmax activation function to output class probabilities.\n\nThis CNN is designed for image classification, particularly for the MNIST digit dataset. 📊💻","metadata":{"id":"ogffQV0oH9Xp"}},{"cell_type":"code","source":"# Build the Convolutional Neural Network (CNN) model using Keras\nmodel = keras.Sequential([\n # First convolutional layer: 32 filters, 3x3 kernel, ReLU activation\n layers.Conv2D(32, (3, 3), activation='relu', input_shape=(28, 28, 1)),\n\n # Max pooling layer: 2x2 pool size\n layers.MaxPooling2D((2, 2)),\n\n # Second convolutional layer: 64 filters, 3x3 kernel, ReLU activation\n layers.Conv2D(64, (3, 3), activation='relu'),\n\n # Max pooling layer: 2x2 pool size\n layers.MaxPooling2D((2, 2)),\n\n # Third convolutional layer: 64 filters, 3x3 kernel, ReLU activation\n layers.Conv2D(64, (3, 3), activation='relu'),\n\n # Flatten the output to feed it into the fully connected layers\n layers.Flatten(),\n\n # Fully connected layer with 64 neurons and ReLU activation\n layers.Dense(64, activation='relu'),\n\n # Output layer with 10 neurons (one for each class) and softmax activation\n layers.Dense(10, activation='softmax')\n])","metadata":{"id":"UyUDS-Ooxc0x","trusted":true,"execution":{"iopub.status.busy":"2025-02-13T23:43:44.627062Z","iopub.execute_input":"2025-02-13T23:43:44.627629Z","iopub.status.idle":"2025-02-13T23:43:44.763159Z","shell.execute_reply.started":"2025-02-13T23:43:44.627595Z","shell.execute_reply":"2025-02-13T23:43:44.762237Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
\n\n# Model Compilation, Training, and Evaluation 📈\n\n1. **Model Compilation**:\n - The model is compiled using the **Adam optimizer**, which is widely used for its efficiency in training deep learning models.\n - The **loss function** used is `sparse_categorical_crossentropy`, suitable for multi-class classification where labels are integers.\n - The **accuracy metric** is used to track the performance of the model during training and evaluation.\n\n2. **Model Training**:\n - The model is trained using the `fit()` method for 5 epochs. During training, it is validated using the test dataset `(x_test, y_test)`.\n\n3. **Model Evaluation**:\n - After training, the model is evaluated on the test dataset using the `evaluate()` method. This returns the test loss and test accuracy.\n - The test accuracy is printed to evaluate the model's performance on unseen data.\n\nThis process trains the CNN on the MNIST dataset and provides a measure of its accuracy on the test set. 🚀🎯","metadata":{"id":"MWaj1nyZIIkw"}},{"cell_type":"code","source":"# Compile the CNN model with Adam optimizer, sparse categorical crossentropy loss, and accuracy metric\nmodel.compile(optimizer='adam',\n loss='sparse_categorical_crossentropy',\n metrics=['accuracy'])\n\n# Train the model using the training data and validate it with the test data\nmodel.fit(x_train, y_train, epochs=5, validation_data=(x_test, y_test))\n\n# Evaluate the model on the test data and obtain test loss and accuracy\ntest_loss, test_acc = model.evaluate(x_test, y_test, verbose=2)\n\n# Print the test accuracy after evaluation\nprint(f'\\n Test ACC : {test_acc}')","metadata":{"id":"sQvraL7Ey656","outputId":"c71f7f4b-2112-4702-c57d-dc138099b118","trusted":true,"execution":{"iopub.status.busy":"2025-02-13T23:43:44.764436Z","iopub.execute_input":"2025-02-13T23:43:44.764709Z","iopub.status.idle":"2025-02-13T23:46:44.876634Z","shell.execute_reply.started":"2025-02-13T23:43:44.764687Z","shell.execute_reply":"2025-02-13T23:46:44.875452Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
\n\n\n# Displaying Sample MNIST Images with Labels 🖼️📊\n\nThis function visualizes a grid of sample images from the MNIST dataset, displaying them alongside their true labels:\n\n### Key Steps:\n1. **Grid Layout**:\n - The grid layout is defined with a fixed number of columns (`cols=5`), and the number of rows is dynamically calculated based on the number of images to display (`num_images`).\n\n2. **Image Display**:\n - Each image is reshaped to its original dimensions (28x28 pixels) and displayed using `plt.imshow()` with a grayscale color map (`cmap='gray'`).\n\n3. **Labels**:\n - Each image is labeled with its corresponding class using `plt.title()`.\n\n4. **Plot Display**:\n - The images are arranged in a grid, and the axes are hidden to focus on the visual content of the images.\n\nThis function is useful for quickly inspecting a subset of images from the dataset. 🎨🔍","metadata":{"id":"Us0P3RGWI20s"}},{"cell_type":"code","source":"import matplotlib.pyplot as plt\n\n# Function to display a grid of images with their corresponding labels\ndef show(images, labels, num_images=10):\n # Define number of columns and calculate number of rows for the subplot grid\n cols = 5\n rows = (num_images // cols) + 1\n\n # Create a figure to display the images\n plt.figure(figsize=(12, 12))\n\n # Loop through the images and plot them on the grid\n for i in range(num_images):\n plt.subplot(rows, cols, i + 1) # Create a subplot for each image\n plt.imshow(images[i].reshape(28, 28), cmap='gray') # Reshape and display the image\n plt.title(f'label:{labels[i]}') # Set the title to the label of the image\n plt.axis('off') # Turn off the axis to focus on the image\n\n # Show the plot with all images\n plt.show()","metadata":{"id":"bugd6CjBP11_","trusted":true,"execution":{"iopub.status.busy":"2025-02-13T23:46:44.877731Z","iopub.execute_input":"2025-02-13T23:46:44.878147Z","iopub.status.idle":"2025-02-13T23:46:44.884639Z","shell.execute_reply.started":"2025-02-13T23:46:44.878079Z","shell.execute_reply":"2025-02-13T23:46:44.883563Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
\n\n\n# Displaying Random Sample Images from Training and Test Datasets 🎲🖼️\n\nIn this cell, we randomly select a subset of images from both the training and test datasets, then display them along with their labels:\n\n### Key Steps:\n1. **Random Index Selection**:\n - `np.random.choice()` is used to randomly select indices from the training (`x_train`) and test (`x_test`) datasets. We select 10 images from the training set and 5 from the test set, without replacement (`replace=False`).\n\n2. **Concatenate the Samples**:\n - The randomly selected images and their labels are concatenated from both datasets into `sample_images` and `sample_labels` using `np.concatenate()`.\n\n3. **Image Display**:\n - The `show()` function is then called to display the selected images and their corresponding labels in a grid layout.\n\nThis provides a quick way to visually inspect random samples from both the training and test datasets. 🎨🔍","metadata":{"id":"fQICZaE4JFsq"}},{"cell_type":"code","source":"import numpy as np\n\n# Select random indices from the training and test datasets\ntrain_indices = np.random.choice(len(x_train), 10, replace=False) # 10 random indices from the training set\ntest_indices = np.random.choice(len(x_test), 5, replace=False) # 5 random indices from the test set\n\n# Concatenate the selected sample images and labels from both training and test datasets\nsample_images = np.concatenate((x_train[train_indices], x_test[test_indices]))\nsample_labels = np.concatenate((y_train[train_indices], y_test[test_indices]))\n\n# Display the sample images with their corresponding labels\nshow(sample_images, sample_labels, num_images=15)","metadata":{"id":"MKbhTJDPTuQv","outputId":"a0a96e64-f655-4e0e-f2df-d996a0387533","trusted":true,"execution":{"iopub.status.busy":"2025-02-13T23:46:44.885736Z","iopub.execute_input":"2025-02-13T23:46:44.886043Z","iopub.status.idle":"2025-02-13T23:46:46.285489Z","shell.execute_reply.started":"2025-02-13T23:46:44.886017Z","shell.execute_reply":"2025-02-13T23:46:46.284236Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
\n\n# Image Upload and Prediction for MNIST Dataset 🖼️🔮\n\nIn this notebook, we allow users to upload an image of a handwritten digit or provide a path to the image. The image is then processed, and the model predicts the digit with confidence. Here's how the system works:\n\n### Key Steps:\n1. **Image Upload**:\n - Users can upload an image of a handwritten digit via the file upload widget (`upload_button`), or provide a file path directly through the `path_input` text box.\n\n2. **Image Preprocessing**:\n - The uploaded image is resized to 28x28 pixels and converted to grayscale. The pixel values are inverted and normalized to a range between 0 and 1 for the model input.\n\n3. **Prediction**:\n - The preprocessed image is passed to the trained model, which predicts the digit. The prediction, along with the confidence level, is displayed alongside the image.\n\n4. **Widgets**:\n - The interactive widgets (file upload, path input, and prediction button) make the process user-friendly.\n\nThis provides an interactive way to input an image, get a prediction, and visualize the result. 🚀💡","metadata":{"id":"Ocqm0SMeJYWV"}},{"cell_type":"code","source":"from PIL import Image\nfrom google.colab import files\nimport ipywidgets as widgets\nfrom IPython.display import display\nimport numpy as np\nimport matplotlib.pyplot as plt\n\n# Define the function to handle file uploads\ndef handle_upload(change):\n # Get the uploaded file and save it as 'user_image.png'\n input_path = next(iter(upload_button.value))\n with open(\"/tmp/user_image.png\", \"wb\") as f:\n f.write(upload_button.value[input_path]['content'])\n path_label.value = \"/tmp/user_image.png\"\n print(\"UPLOAD DONE\")\n\n# Define the function to handle path input\ndef handle_path_input(change):\n path_label.value = path_input.value # Update path_label with the input path\n print(f'PATH DONE ENTERED')\n\n# Define the function to handle the prediction\ndef handle_prediction(change):\n # Get the image path from the label\n input_path = path_label.value\n if not input_path:\n print(\" input invalid \")\n return\n\n # Load and preprocess the image\n img = Image.open(input_path).convert('L')\n img = img.resize((28, 28)) # Resize to 28x28\n img_array = np.array(img)\n\n # Invert the colors and normalize the image\n img_array = 255 - img_array\n img_array = img_array / 255.0\n img_array = img_array.reshape(1, 28, 28, 1)\n\n # Make the prediction\n prediction = model.predict(img_array)\n prediction_label = np.argmax(prediction) # Get the predicted label\n confidence = np.max(prediction) * 100 # Get the prediction confidence\n\n # Display the image and prediction result\n plt.imshow(img, cmap='gray')\n plt.title(f'predict:{prediction_label} (Confidence: {confidence:.2f}%)')\n plt.axis('off')\n plt.show()\n\n print(f'Predicted digit: {prediction_label} with {confidence:.2f}% confidence.')\n\n# Create the necessary widgets\nupload_button = widgets.FileUpload(accept='image/*', multiple=False) # Upload button\npath_input = widgets.Text(description=\"Path to image:\") # Text input for image path\nsubmit_button = widgets.Button(description=\"Click To Predict\", style={'button_color': 'green', 'font_weight': 'bold'}) # Button to trigger prediction\npath_label = widgets.Label(value=\"\") # Label to show the image path\n\ndef predict_input():\n \"\"\"\n This function initializes the widgets and displays them. It also binds the\n handle_upload, handle_path_input, and handle_prediction functions to the\n corresponding widgets.\n \"\"\"\n # Bind the functions to the respective widgets\n upload_button.observe(handle_upload, names='value')\n path_input.observe(handle_path_input, names='value')\n submit_button.on_click(handle_prediction)\n\n # Display the widgets\n display(widgets.Label(\"Choose how to input your image:\"))\n display(widgets.HBox([upload_button, path_input])) # Upload button and path input side by side\n display(path_label) # Display the path label\n display(submit_button) # Display the submit button\n\n# Initialize the widgets and display them\npredict_input()\n","metadata":{"id":"4eUFeLph1nnP","outputId":"e40c028f-fbd2-4296-ba4d-3f20bef42bae","trusted":true,"execution":{"iopub.status.busy":"2025-02-13T23:46:46.286823Z","iopub.execute_input":"2025-02-13T23:46:46.287263Z","iopub.status.idle":"2025-02-13T23:46:46.344290Z","shell.execute_reply.started":"2025-02-13T23:46:46.287207Z","shell.execute_reply":"2025-02-13T23:46:46.343101Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
\n
\n
![\"Typing](\"https://readme-typing-svg.demolab.com?font=Fira+Code&size=25%&pause=1000&color=00abf0¢er=true&vCenter=true&width=435&height=25%&lines=Thank+you+for+visiting!;Have+a+good+day!;\")
\n
\n
![](\"https://capsule-render.vercel.app/api?type=waving&color=00abf0&height=70§ion=footer\"/)
","metadata":{"id":"4OlNJXP8QOh8"}}]}