This repository contains the implementation of a custom Convolutional Neural Network (CNN) to classify images from the FashionMNIST dataset. The model is designed to showcase advanced features like data augmentation, regularization, and dynamic hyperparameter tuning.
- Custom CNN Architecture: Includes multiple convolutional layers, batch normalization, dropout regularization, and fully connected layers.
- Data Augmentation: Random rotations and horizontal flips to improve model generalization.
- Dynamic Hyperparameters: Adjustable learning rate, batch size, dropout, and number of epochs via command-line arguments.
- Visualization: Generates confusion matrices and displays sample predictions for better model interpretation.
- Performance Tracking: Logs training and validation accuracy to identify overfitting or underfitting.
Clone this repository to your local machine:
git clone https://github.com/cxycode32/CustomCNN-FashionMNIST.git
cd CustomCNN-FashionMNISTInstall the required dependencies:
pip install -r requirements.txtRun the training script with default parameters:
python main.pyCustomize hyperparameters using command-line arguments:
python main.py --learning_rate 0.0005 --batch_size 32 --num_epochs 20 --dropout 0.4├── assets/ # Images of accuracy plot, confusion matrix, and test result
├── main.py # Main script for training and evaluation
├── README.md # Project documentation
└── requirements.txt # Required dependencies
- Final Test Accuracy: The model achieves competitive performance on the FashionMNIST dataset.
- Confusion Matrix: Provides detailed insights into classification performance for each class.
The accuracy of the model during training and validation is plotted to monitor progress:
Feel free to fork this repository and submit pull requests to improve the project or add new features.
This project is licensed under the MIT License.