TinyVGG: Optimized Image Classification Model

Try live at: https://tinyvgg.streamlit.app

📘 Overview

TinyVGG is a lightweight image classification model inspired by the VGG16 architecture, re-engineered for efficiency, portability, and high performance. It achieves 92% classification accuracy on the CIFAR-10 dataset while maintaining a tiny model size of just 4MB, making it ideal for deployment on resource-constrained devices such as mobile and embedded systems.

🚀 Features

• High Accuracy: Achieved 92% test accuracy on CIFAR-10, matching larger models with fewer parameters.
• Compact Model: Reduced from ~528MB (original VGG16) to 4MB without major performance loss.
• Optimized for Deployment: Lightweight architecture designed for real-time inference and low-latency predictions.
• Preprocessing Pipeline: Built-in image normalization, resizing, and augmentation for consistent input quality.
• Interactive Demo: Easily test model predictions via a Streamlit web interface.

🧪 Dataset

Dataset: CIFAR-10

Classes: Airplane, Automobile, Bird, Cat, Deer, Dog, Frog, Horse, Ship, Truck

Size: 60,000 images (32×32 pixels, RGB)