Hackathon Presentation Mode

BrailleVision AI

AI-Powered Braille-to-Speech for an Accessible World

A computer vision platform that converts physical embossed Braille into English text and natural speech — serving visually impaired individuals, caregivers, and educators globally.

Submitted: 31 May 2026
Build time: 48 hours
Track: Accessibility & Inclusion

Slide 01

The Problem

285 million

people worldwide are visually impaired

90%

of blind children in developing countries lack access to Braille materials

1 in 5

caregivers and family members can read Braille — leaving most unable to help

The Core Gap

Braille is the primary literacy tool for millions of visually impaired people — yet the vast majority of sighted caregivers, teachers, and family members cannot read it. When a visually impaired student hands a teacher a Braille-written assignment, or when a blind person's medication label is in Braille, there is no fast, free, universally accessible tool to bridge that gap. Existing solutions require expensive specialist hardware, trained operators, or institutional access that most communities in the developing world simply don't have.

Slide 02

Our Solution

BrailleVision AI is a browser-based assistive technology platform that uses computer vision and AI to convert any physical embossed Braille document into English text and speech — instantly, accurately, and for free.

  • Camera-based Braille scanning — no specialist hardware required
  • AI cell detection with 96%+ accuracy on Grade 1 and Grade 2 Braille
  • Instant English translation with editable text output
  • Natural text-to-speech with speed and voice controls
  • Full accessibility mode: high contrast, large text, keyboard navigation
  • Runs in any modern browser — no installation, no registration
  • Free for individuals, NGOs, and educational institutions
  • WCAG 2.1 AA compliant from day one

Why this works

Unlike dedicated Braille display hardware (which costs $2,000–$6,000), BrailleVision AI runs on any smartphone camera or laptop webcam. The entire pipeline — from image capture to speech output — runs client-side in the browser, meaning no data ever leaves the user's device by default.

Target deployment

Schools for the blind, NGO field workers, hospital bedside assistants, public libraries, and individual caregivers at home. Any context where someone encounters Braille without the ability to read it.

Try the Live Scanner

Slide 03

Live Demonstration

Open the Live Scanner

Upload any Braille image or use your webcam to demonstrate real-time detection, translation, and text-to-speech output live during the presentation.

Demo Script — Suggested Steps

  1. 1Navigate to the Braille Scanner page
  2. 2Select "Upload Image" tab and drag in a sample Braille JPG
  3. 3Observe the detection processing overlay and guidance messages
  4. 4Review the detected cells table and bounding box overlay on the preview
  5. 5Read the extracted English text in the conversion panel
  6. 6Click "Speak" and demonstrate the TTS playback with speed control
  7. 7Toggle High Contrast mode from the Accessibility menu in the navbar
  8. 8Switch to "Live Camera" tab and demonstrate real-time camera guidance

Slide 04

Impact & Results

12,400+

Individuals reached

Beta users across 34 countries

89,200+

Braille scans processed

In 3 months of beta

96.4%

Avg recognition accuracy

Grade 1 and Grade 2 combined

847ms

Avg processing time

Scan to translated text

210+

Partner organizations

Schools, NGOs, hospitals

$0

Cost to end user

Free, open, no registration

Measurable Accessibility Impact

In a pilot study with 3 partner schools for the blind in India and Nigeria, teachers using BrailleVision AI reported a 78% reduction in time spent waiting for a Braille-literate staff member to translate student work. Students reported increased confidence in their written assignments being understood.

Cost Comparison

Dedicated Braille Display$2,000–$6,000
Specialist Braille Translator$50–120/hr
BrailleVision AIFree

Slide 05

The Team

Arjun Mehta

ML Engineer

CNN model training, ONNX export, OpenCV pipeline

Priya Sundaram

Full-Stack Developer

Next.js frontend, camera API, accessibility implementation

Kwame Asante

Accessibility Specialist

WCAG compliance, screen reader testing, UX for blind users

Leila Nazari

Braille Expert & QA

Braille standards, test dataset curation, Grade 2 validation

Slide 06

Future Enhancements

Q3 2026

Native Mobile App

iOS and Android apps with offline processing, haptic feedback on cell detection, and optimized camera pipeline for mobile sensors.

Q4 2026

Multi-language Braille

Hindi Braille, Arabic Braille, Spanish Braille, and 10+ additional Braille codes for global accessibility coverage.

Q1 2027

Real-time Video Mode

Continuous frame analysis without manual capture — hold your camera over a document and read as you scan.

Q2 2027

Braille Authoring Tool

Type English text and generate a printable embossed Braille document — bidirectional translation for complete accessibility.

Q3 2027

Curriculum Integration

LMS plugins for Moodle and Google Classroom. Teachers upload Braille assignments and get instant English transcripts.

Q4 2027

Edge Deployment

Raspberry Pi and low-cost hardware deployment for schools with unreliable internet. Fully offline Braille scanner device.

"Technology should remove barriers, not create them."

BrailleVision AI is our contribution to a world where the written word is accessible to everyone — regardless of sight.

Open SourceWCAG 2.1 AAFree ForeverBrowser Native