ReCallMe

ReCallMe - Empowering Memory, One Moment at a Time. A wearable companion that not only supports dementia patients in their daily lives but actively engages them in recalling precious memories

Comment

RecallMe: AI-powered memory assistant for dementia patients

Inspiration

The challenge of caring for individuals with dementia affects millions of families worldwide. We were inspired by personal experiences with family members struggling with memory loss and the daily challenges they face. Current solutions often focus on basic reminders or tracking, but we envisioned a more comprehensive, real-time assistant that could help patients understand their surroundings and maintain independence longer and train their memory with highly-tailored quizzes.

What it does

RecallMe is an AI-powered assistant that serves as a real-time memory aid and therapeutic platform for people with dementia. Key features include:

Real-time Assistance:

  • Uses advanced computer vision to identify objects, people, and potential hazards in the environment
  • Provides clear, simple descriptions of surroundings through both visual and audio feedback
  • Actively monitors for safety concerns and offers gentle guidance
  • Records important moments and interactions throughout the day

Memory Training Platform:

  • Creates personalized memory exercises using content gathered during daily activities
  • Converts real-world experiences into interactive memory training sessions
  • Helps maintain cognitive function through engagement with familiar content
  • Provides structured activities for patient-caregiver interaction

How we built it

We developed ReCallMe using cutting-edge AI technologies and efficient architecture:

Core Technologies:

  • Computer Vision: YOLOv11 for real-time object detection (30+ FPS)
  • AI Understanding: Claude 3 Haiku and Gemini 1.5 pro for context-aware scene descriptions
  • Multi-threaded Processing: Parallel handling of video, AI inference, and user interface
  • Efficient Data Management: Custom queue system for real-time processing

Technical Implementation:

  • Python-based architecture with OpenCV for video processing
  • Multi-threaded design for smooth performance
  • Optimized resolution and frame selection for real-time processing
  • Efficient memory management for continuous operation
  • User friendly and extremely simple interface with React

Challenges we ran into

Real-time Performance:

  • Balancing AI processing speed with accuracy
  • Managing multiple AI systems simultaneously
  • Optimizing memory usage for continuous operation

Accomplishments that we're proud of

  1. Technical Achievements:

    • Real-time object detection and scene understanding
    • Smooth integration of multiple AI systems
    • Efficient multi-threaded architecture

  2. User-Focused Design:

    • Intuitive, non-overwhelming interface
    • Clear, actionable descriptions
    • Therapeutic memory training integration

  3. Impact:

    • Comprehensive support for both daily assistance and cognitive maintenance
    • Scalable platform for future enhancements
    • Assistive + therapeutic technology

What we learned

  • Advanced techniques in real-time AI system integration
  • Importance of balancing technical capability with user accessibility
  • Methods for making AI communications more human-friendly
  • Techniques for efficient multi-threaded application development

What's next for RecallMe

Technical Enhancements:

  • Voice command integration
  • Edge computing support

Feature Expansion:

  • Location awareness for navigation
  • Emergency contact system
  • Activity recognition

Memory Training Expansion:

  • User homepage with other stats from health-tracking devices

ReCallMe aims to revolutionize dementia care by providing both immediate assistance and long-term cognitive support. Our combination of real-time aid and therapeutic training creates a comprehensive solution for maintaining independence and quality of life.

Share this project:

Updates