CIFAR-10 Classification

This project explores image classification on the CIFAR-10 dataset using a custom neural network architecture.

Results

The implemented model achieves 85% test accuracy on CIFAR-10.

For comparison, a baseline ResNet architecture reaches 88% accuracy, resulting in only a 3% performance gap.

Dataset

CIFAR-10 is a standard benchmark dataset consisting of 60,000 32×32 color images across 10 classes.

Technologies

• Python
• PyTorch
• NumPy