EMNIST Digit Prediction Web Application

Description:

A web application that takes a user drawn image and uses a convolutional neural network to recognize the image's digit classification.

Convolutional Neural Network

The bulk of this project is the Java Springboot backend, which is a hand-rolled convolutional neural network for EMNIST digit classification. The neural network can train, test, and then with an accuracy above 90% predict the drawn images.

EMNIST Data

The EMNIST dataset is a set of handwritten digits split into 60,000 training and 10,000 testing images.

---

Application Demo:

Show

Screenshots:

Show

---

Features:

Frontend:

• Vue
• Vite
• Tailwind
• Formatting (Prettier)

Backend:

• Java
• Maven
• SpringBoot
• Parquet (EMNIST Data)

---

Developer Notes:

# PACKAGE THE APPLICATION
mvn clean package

# PACKAGE THE APPLICATION (SKIP TESTS)
mvn clean package -DskipTests

# RUN APPLICATION ON (http://localhost:8080)
mvn spring-boot:run &

---

Resources Used:

Source	Description
Part One and Part Two	Tutorial by Victor Zhou on convolutional neural networks
Blog	Explanation of convolutional neural networks
frontend-maven-plugin	Helps integrate the frontend and backend

The beginning of wisdom is: Acquire wisdom; And with all your acquiring, get understanding.