About This Project
Overview
A PyTorch and OpenCV image classification pipeline trained to categorize fashion items in the Fashion MNIST dataset — a foundational deep-learning exercise in convolutional classification.
Motivation
This was a foundational exercise to build convolutional classification intuition from the ground up before tackling harder computer-vision problems.
Lessons Learned
Even on a 'solved' dataset like Fashion MNIST, small architecture and augmentation choices had a surprisingly large effect on accuracy — a good reminder that fundamentals matter even on toy problems.