Be My Eyes - Leveraging Generative AI to help the blind and visually impaired people by Danushka

Ai

Learn how Generative AI & edge computing can create affordable personal assistant devices for the visually impaired, making daily navigation & tasks more accessible.

Key takeaways
  • Project aims to build an affordable personal assistant device for blind and visually impaired people using Generative AI and edge computing

  • Device requirements:

    • Must be portable/wearable with self-contained power
    • Needs camera input and text-to-speech capabilities
    • Should have minimal, touch-friendly interface
    • Must be affordable since many blind people have limited financial resources
  • Technical implementation options:

    • Raspberry Pi-based solution with local LLMs
    • Smartphone-based application leveraging built-in capabilities
    • Integration with streaming platforms like Red Panda/Kafka for scalability
  • Key features:

    • Real-time image/video description
    • Natural language interaction
    • Hazard detection and warnings
    • Geofencing and location awareness
    • Integration with medical monitoring systems
  • Current challenges:

    • Running large vision-capable LLMs on edge devices
    • Balancing performance with cost
    • Internet connectivity requirements
    • Model optimization for resource-constrained environments
  • Future potential applications:

    • AI-assisted body cameras for rescue workers
    • Remote patient monitoring
    • Assistance for elderly care
    • Personal navigation support
  • Focus on social inclusion through affordable, practical technology rather than complex research solutions

  • Project emphasizes sustainability through open-source components and local processing where possible

  • Integration capabilities with healthcare systems and monitoring solutions for comprehensive care

  • Design prioritizes enhanced audio cognition capabilities of blind users and natural interaction patterns