AI-Powered Accessibility

Assistance for the Visually Impaired

Real-time AI-powered tools that help visually impaired individuals understand and navigate the world around them.

Bridging the Vision Gap with Deep Learning

Understanding the world without sight presents unique challenges, from recognising objects to navigating unfamiliar spaces. This AI-powered assistive application helps visually impaired individuals perform everyday visual tasks with confidence.

Built on the VizWiz dataset, real-world images captured by blind individuals, the system uses ResNet50 and Bi-LSTM architectures to deliver accurate image captioning. The mobile app, developed in React Native, prioritises privacy and security by keeping all image processing on-device without third-party APIs.

ResNet50 Bi-LSTM React Native VizWiz Dataset On-Device AI

Key Features

A comprehensive set of AI-driven tools designed to make visual information accessible to everyone.

Image Captioning

Generates detailed, natural-language descriptions of images in real time to convey visual context.

Location-Based Tasks

Provides context-aware assistance tailored to the user's surroundings and environment.

Image Quality Feedback

Detects blur, overexposure, and underexposure to guide users toward better image capture.

Text & QR Code Detection

Reads signs, labels, and documents so users can access printed information independently.

Obstacle Detection

Identifies obstacles in the user's path to enhance navigation safety in real time.