SignLook

SignLook: Transform hand gestures into text in real-time using your computer's camera. A tool for learning sign language and bridging communication gaps.

Comment

Inspiration

When my best friend recently lost his hearing, our usual coffee meetups became challenging. I found myself struggling to communicate effectively, and watching him navigate this new reality was tough. This personal challenge sparked the idea for SignLook - I wanted to create something that could help bridge the communication gap while I learned sign language myself.

What it does

SignLook is a real-time sign language recognition web app that:

  • Converts hand gestures into text instantly using your computer's camera
  • Achieves 91% accuracy in letter recognition
  • Allows manual letter confirmation with the Enter key
  • Includes prediction stabilization for reliable detection
  • Lets you save and view your typed phrases
  • Features a modern, user-friendly interface

How I built it

I built SignLook entirely solo, diving deep into:

  • Computer vision and machine learning for hand gesture recognition
  • Flask for the backend web server
  • Real-time video processing using OpenCV
  • A responsive frontend with JavaScript
  • Creating a dataset of over 1000 hand gesture images
  • Training a machine learning model that achieves 91% accuracy

Challenges I ran into

The biggest challenge was doing this entire hackathon during a massive blackout! I was literally running between libraries and cafés, chasing Wi-Fi signals to keep working. Other challenges included:

  • Creating a diverse dataset of hand gestures
  • Dealing with varying lighting conditions
  • Ensuring real-time performance
  • Making the interface intuitive for first-time users
  • Learning sign language while building the app

There were moments when I thought I'd have to give up, but I refused to let a power outage stop me from creating something meaningful for my friend.

Accomplishments that Im proud of

  • Built a fully functional sign language recognition system from scratch
  • Created my own dataset of over 2000 hand gesture images using scripts
  • Achieved 91% accuracy in letter recognition
  • Developed an intuitive, user-friendly interface
  • Learned sign language while building the app
  • Completed the project entirely solo during a blackout
  • Successfully surprised my friend with the working prototype

What I learned

This project taught me so much, both technically and personally:

  • Computer vision and machine learning fundamentals
  • Real-time video processing techniques
  • Web development with Flask
  • UI/UX design principles
  • Basic sign language (Entire alphabet and basic words like hello etc.)
  • The importance of perseverance
  • How technology can truly make a difference in people's lives

What's next for SignLook

I'm constantly working to improve SignLook. Future plans include:

  • Adding support for full words and phrases
  • Improving accuracy in different lighting conditions
  • Creating a mobile app version
  • Adding a learning mode for beginners
  • Supporting multiple sign language dialects
  • Building a community of users and contributors

I'd love your feedback and contributions to make SignLook even better! Whether you're learning sign language, know someone who's hearing impaired, or just interested in the technology, I hope SignLook can make a difference in your life too.

Built With

Share this project:

Updates