• 

  Our Card

• 

  Accuracy Rates

Asciende

Asciende is an interactive web app that recognizes American Sign Language (ASL) letters in real time and provides instant feedback on how to perfect each sign.
Built with Streamlit, OpenCV, and a custom-trained model, the app also displays a reference image to help users improve their form.

Features

• Live Webcam Recognition – Detects ASL letters continuously from your camera feed.
• Correction Feedback – For every letter you sign, the app shows:
  • Text guidance on how to adjust your hand shape.
  • A sample image from our reference dataset for comparison.
• Streamlit UI – Clean, responsive interface that runs entirely in the browser.

How It Works

1. Webcam Capture – OpenCV streams frames directly from your webcam.
   
2. Prediction – A trained machine learning model identifies the current ASL letter.
   
3. Feedback – The app provides:
   
   • Text instructions on how to improve the sign.
     
   • A sample image from asl_dataset/<Letter>/ for visual guidance.
     

Inspiration & Learnings

This project was inspired by the desire to make ASL learning more accessible.

While building Asciende, we learned about:

• Real-time image processing pipelines.
  
• Training and serving custom computer vision models.
  
• Streamlit’s powerful yet simple UI components.
  

Next Steps

• Add support for full words/phrases.
• Instead of Custom Letters, use offical ASL
• Improve accuracy with additional training data.
  

How to Run:

Built With

• python
• streamlit
• teachablemachine