Building an Accessible Web App for the Visually Impaired — My Journey, Inspiration & Vision 👁️‍🗨️

“Technology should empower everyone, especially those who need it the most.”

🌟 The Idea

While most developers showcase cool AI demos — hand gestures to play games, facial filters, or fancy object detection — I kept asking myself: “Who’s actually benefiting from this?”

That’s where my idea was born:
A browser-based assistive tool for visually impaired people that uses real-time object detection, audio feedback, and possibly haptic feedback to help navigate the world around them.

👓 The Inspiration

In college (DU), I came across a professor who is blind. Watching how he engaged with tech through accessibility tools — often limited or clunky — made me think:

"What if I could use the same tools we developers play around with, but build something that could actually assist someone like him in day-to-day life?"

That’s where it all started.

🔍 The Core Vision

This app will work in the browser and allow the user to:

• Use their device camera to scan their surroundings
• Detect objects in real-time (cars, people, potholes, signs, etc.)
• Provide audio feedback (using Web Speech API)
• Optionally provide vibration/haptic patterns for non-verbal alerts
• Ask users during setup if they prefer voice and/or vibration feedback

All with a focus on simplicity, accessibility, and ease-of-use.

🛠️ Tech Stack

Here’s what I’m planning to use:

🧪 Future Features

• Pothole detection (custom-trained model on Indian roads)
• QR code recognition for reading signs/info
• Face/person detection with alerts
• Offline support via PWA capabilities
• Integrating voice-based interaction to operate the app hands-free

💭 Why Not Just Build It With OpenCV?

Yes, OpenCV is powerful, but most of these projects:

• Require local setup
• Are demo-heavy, not user-focused
• Aren’t made for real people who aren’t tech-savvy

My goal?
Build something that just works in a browser — no setup, no installs.
If someone can open Google Chrome, they should be able to use this app.

🧠 Lessons I Plan to Learn Along the Way

• TensorFlow.js for real-time detection
• Accessibility-first UI/UX
• Using browser APIs in production
• Deploying models and optimizing for performance
• Possibly training a custom model for potholes or Indian-specific objects

🤝 Call to Collaborate

If you're also working on accessibility, browser-based ML, or open-source assistive tech, I'd love to connect or collaborate.
Let’s make tech actually helpful for those who need it most.

Connect with me on LinkedIn: @sidharthsangelia

Thanks for reading. Let’s build for impact. 💙