CIFAR-10 Image Classification with CNN and Batch Normalization This project implements a convolutional neural network (CNN) with batch normalization for image classification on the CIFAR-10 dataset. It utilizes PyTorch libraries and techniques for efficient training and evaluation.

Project Highlights:

Achieved a validation accuracy of 88.2% after 10 epochs of training. Employed batch normalization for better performance and reduced overfitting. Leveraged efficient PyTorch libraries for training and evaluation. Installation:

This project requires the following libraries:

PyTorch torchvision numpy matplotlib Install the required libraries using pip:

pip install torch torchvision numpy matplotlib Getting Started:

Download the CIFAR-10 dataset: python -m datasets download cifar10 Run the script: python main.py Code Structure:

main.py: The main script that defines the model, training loop, and evaluation process. ImageClassificationBase.py: A base class for image classification models with common training and validation steps. CIFAR_CNN.py: The CNN model architecture specifically designed for the CIFAR-10 dataset. utils.py: Utility functions for device allocation and data loading. Project Contributions:

This project is a learning exercise for implementing CNNs with batch normalization for image classification. It serves as a baseline model for further experimentation and improvement. The code is well-structured and documented for easy understanding and modification. Future Work:

Explore different CNN architectures and hyperparameters for improved performance. Implement data augmentation techniques to increase training data diversity. Evaluate the model on other image classification datasets.