Building RecallMe: An On-Device AI Companion for Dementia Care Using Flutter & Kiro

Ever thought about building an AI app that runs entirely on a phone, no internet required? The "RecallMe" project, built with Flutter and accelerated by the AI-powered IDE Kiro, shows us how. This app...

Building RecallMe: An On-Device AI Companion for Dementia Care Using Flutter & Kiro
Photo by Artur Shamsutdinov / Unsplash

Ever thought about building an AI app that runs entirely on a phone, no internet required? The "RecallMe" project, built with Flutter and accelerated by the AI-powered IDE Kiro, shows us how. This app is designed to help people with dementia recognize faces and recall memories, addressing a critical need with a privacy-first approach. The heart of the app is its on-device face recognition, powered by Google ML Kit and a custom-built embedding algorithm in Kotlin, optimized for Arm processors. This algorithm combines color histograms, spatial grid features, and texture analysis to create a 256-dimensional vector for each face, enabling quick and accurate matching, and ensuring the app runs smoothly even on mid-range devices.

The magic behind RecallMe wasn't just the tech stack, but also the development process. By using Kiro, the developer was able to rapidly prototype and build features through "vibe coding." This involved creating steering documents that outlined the project's architecture, design principles, and tech stack, allowing Kiro to generate code that seamlessly fit the existing codebase. Imagine simply telling your IDE, "Build a routine management screen with add/edit/delete functionality," and it generates a complete implementation, complete with state management using Provider and data persistence using Hive – all while adhering to a dementia-friendly color theme! This approach drastically reduced development time, allowing the team to focus on refining the app's features and ensuring its accessibility.

RecallMe showcases a number of powerful features, including real-time face recognition, memory recall triggered by photos, routine management with notifications, and voice interaction. Crucially, all data stays on the device, ensuring privacy. The source code is available on GitHub, offering a treasure trove of insights into on-device ML techniques, dementia-friendly design, and Arm architecture optimization. This project serves as an inspiration and a practical guide for developers looking to build AI applications that are both powerful and respectful of user privacy. Check out the project, `git clone https://github.com/Gaurav-derry/Recall`, and see how you can contribute to building a better future with AI.


📰 Original article: https://dev.to/gaurav_pandey_a16499e79dc/building-recallme-an-on-device-ai-companion-for-dementia-care-using-flutter-kiro-1kj3

This content has been curated and summarized for Code Crafts readers.