A Simple Digit Classification program made using Convolutional Neural Network

This project involves building a Convolutional Neural Network (CNN) to classify digits. The model is trained using image data representing various digits and utilizes techniques from deep learning and image processing.

The project involves:

• Preprocessing the image data by resizing, converting to grayscale, and applying histogram equalization.
• Implementing a CNN architecture using the Keras deep learning framework.
• Augmenting the training data for better model generalization.
• Training the model on a dataset of digit images.
• Evaluating the trained model's performance on a test dataset.

• Libraries: NumPy, OpenCV (cv2), os, Keras, Matplotlib, and Pickle.
• Data Handling: Utilizing train-test split and generating training, testing, and validation sets.
• Image Preprocessing: Converting images to grayscale, equalizing histograms, and normalization.
• Data Augmentation: Using ImageDataGenerator for augmenting training data.
• Model Architecture: Utilizing Conv2D, MaxPooling2D, Dropout, Dense layers in the CNN.
• Model Compilation: Using Adam optimizer, categorical crossentropy loss, and accuracy metrics.
• Visualization: Plotting bar graphs for image class distributions and loss/accuracy trends.
• Serialization: Saving the trained model using Pickle for future use.

1. Ensure all necessary libraries are installed (Keras, NumPy, OpenCV, Tensorflow, etc.).
2. Set up the dataset in a directory named 'myData' containing subdirectories for each class of digit images.
3. Run the Python script provided to train the model and evaluate its performance.