Memoir

Millions struggle with memory loss, forgetting faces and moments. Memoir uses AI to sort photos, recognize faces, and build relationships—helping you relive and never forget your cherished memories.

Comment

Memoir: A Journey in Preserving Memories

Inspiration

The idea for Memoir was born from a deeply personal realization—how fragile our memories can be. Seeing how Alzheimer’s and memory loss affect individuals and their loved ones, we wanted to create a tool that helps people reconnect with their past. We envisioned an app that could sort, recognize, and retell the moments that define our lives, ensuring that no memory fades into oblivion.

What We Learned

Building Memoir taught us invaluable lessons about machine learning, user experience, and accessibility. We explored facial recognition models, audio listeners, and data organization techniques to make memories easily searchable and narratable. Additionally, we delved into the ethical concerns of data privacy, ensuring that user memories remain secure.

How We Built It

  • Frontend: Developed using React Native for a smooth cross-platform experience.
  • Backend: Powered by Firebase for real-time data storage and authentication.
  • AI Integration: Implemented Face Recognition (Facenet512) to identify people in images/videos.
  • Audio Analysis: Used speech-to-text models (Chirp2) to capture conversations and add context to memories.
  • Summarization: Finally, used Gemini's latest 2.0 flash model to summarize the long conversations into understandable summaries.
  • Memory Sorting: Designed a relationship-based profile system to categorize moments based on people and events.

Challenges We Faced

  • Facial Recognition Accuracy: Training the AI to correctly identify people across different lighting, angles, and aging variations.
  • Multi-Model Integration: Coordinating multiple AI models—face recognition, speech-to-text, and contextual understanding—was complex. Ensuring they communicated efficiently and produced accurate, meaningful outputs required extensive tuning and optimization.
  • User Experience: Making the app intuitive for both tech-savvy users and those who may struggle with cognitive decline.

The Future of Memoir

Our journey with Memoir is just beginning. We aim to integrate voice-activated memory recall, AR storytelling, and collaborative memory-sharing features. Our vision is to make Memoir a digital companion that not only stores moments but also tells the stories that matter. We hope to do this via a integration of meta glasses allowing for real time data to be transmated.

Because memories shouldn’t just be stored—they should be remembered.

Share this project:

Updates