VaibhavUpreti / handwriting-recognition

handwriting recognition flask web app

Handwriting Recognition Web Application

This application is designed to recognize handwriting from images.

Demo

Check out the working demo in action:

handy.mp4

Installation

Choose one of the following methods to install the application:

1. Manual Setup
2. Docker
3. Helm Chart (Kubernetes)

Manual Setup

If you prefer to set up the application manually

Requirements

Make sure you have the following prerequisites installed:

1. Python version >= 3.8
2. Required Python packages can be found in requirements.txt.
3. CtCWordBeamSearch
4. SimpleHTR (optional)

Steps

Recommended to use pyenv 3.10

pip install -r src/requirements.txt

This will install the following packages

• opencv-python
• editdistance
• ultralytics
• flask
• tensorflow
• lmdb
• path
• gunicorn

git clone https://github.com/githubharald/CtCWordBeamSearch
cd CtCWordBeamSearch
pip install .

Start Server

python3 src/app.py

Navigate to http://localhost:3000

Docker

Supported Docker images for your convenience:

1. linux/amd64 - Built using GitHub Actions Runner. See the workflow file for details.
2. linux/arm64 - Available on Docker Hub.

docker pull ghcr.io/vaibhavupreti/handwriting-recognition:latest
docker run -p 3000:3000 <image_id>

Helm Chart (Kubernetes)

Using Helm...

helm install handy helmchart

Tech Stack

1. Flask
2. AWS EC2
3. Caddy web server