MNIST DCGAN

A Deep Convolutional Generative Adversarial Network (DCGAN) trained on the MNIST dataset to generate realistic handwritten digits.

Overview

This project implements a DCGAN — a generator and discriminator pair of convolutional networks trained adversarially — to synthesize handwritten digits that resemble the MNIST training distribution.

Highlights

• Convolutional generator and discriminator architecture
• Adversarial training on the MNIST handwritten-digit dataset
• Interactive demo deployed on Streamlit

Tech Stack

• Language: Python
• Domain: Deep learning / generative models
• Demo: Streamlit

Live Demo

Try it at mnist-aayan.streamlit.app.