TextRecognition

Team

• Ahmad Al-Jabi: LinkedIn, GitHub
• Davi Gava Bittencourt: LinkedIn, Github

Project Overview

TextRecognition is a CNN built during the Winter 2025 edition of MAIS 202. As its name suggest, TextRecognition is able identify all hand-written alphanumerical characters in a word. In terms of dataset, we decided to use a fraction of the NIST dataset complemented by custom data made by us using our own UI.

Dataset: Google Drive

Setting Repo and Installing Dependencies

- Clone repo:

$ git clone https://github.com/Davigbit/TextRecognition

$ cd TextRecognition

- Create an environment and install server dependencies on Linux or Mac:

$ python3 -m venv text-recog-env

$ source text-recog-env/bin/activate

$ pip install -r requirements.txt

- Create an environment and install server dependencies on Windows:

$ python -m venv text-recog-env

$ text-recog-env/Scripts/activate

$ pip install -r requirements.txt

- Install client dependencies:

$ deactivate

$ cd text-recognition-ui

$ npm install --legacy-peer-deps

Deployement

Open two terminals, one for running the server, the other for the client. Also, make sure that both terminals are inside the project's folder.

- Running server on Linux or Mac:

$ source text-recog-env/bin/activate

$ python3 src/server.py

- Running server on Windows:

$ text-recog-env/Scripts/activate

$ python src/server.py

- Running client:

$ cd text-recognition-ui

$ npm run dev

Go to http://localhost:5173/

- Opening notebooks on Linux or Mac:

$ source text-recog-env/bin/activate

$ jupyter notebook

- Opening notebooks on Windows:

$ text-recog-env/Scripts/activate

$ jupyter notebook