ReCall

ReCall is a lightweight memory assistant that turns raw conversations into something you can actually browse. We pair an Expo-based mobile app (“Home”, “Memory”, “Highlights”, “Upload”) with a Flask backend that handles video uploads, face recognition, transcript analysis, and highlight detection.

Quick Tour

• Home – a quick at-a-glance of the constellation of people you've talked to, highlights coming up, and shortcuts to upload or browse memories.
• Memory – every person, their latest transcript, and a scoped “Ask AI” section that only references their conversations.
• Highlights – a short list of reminders (birthdays, next steps, etc.) you can complete or dismiss with a single tap.
• Upload – drop in a new video, we do the rest (faces, transcripts, and linking to the right person).

Running ReCall

# Backend
cd backend
python -m venv .venv && source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -r requirements.txt
cp .env.example .env   # then copy in your AWS/Gemini keys + BASE_URL (see below)
python app.py

# Frontend
cd frontend
npm install
cp config.example.js config.js   # set BASE_URL to your backend
npm run start                    # expo start

Environment keys

Populate backend/.env with the same values used in development:

AWS_ACCESS_KEY_ID=
AWS_SECRET_ACCESS_KEY=
AWS_REGION=
GEMINI_API_KEY=
BASE_URL=

The speech-to-text analyzer also expects backend/analyzers/google_key.json to contain the same Google Cloud service account JSON you used while building the project. Drop that JSON file in place before running app.py.

Use Expo Go (or a simulator) to open the QR code shown in the terminal.

Why This Is Ethical

We spent time documenting how we capture and use memories responsibly. You can read that outline here: Ethical Use of ReCall

In short: recordings must be consensual, we keep memories on your device/back-end only, and every highlight is meant to help with relationship follow‑ups—not surveillance.

Project Structure

backend/      # Flask API, analyzers, services (highlights, LinkedIn enrichment, etc.)
frontend/     # Expo app with Home, Memory, Highlights, Upload screens
docs/images/  # Pitch deck snippets and screenshots

Tech Grab Bag

• Frontend – React Native 0.81 (Expo 54), Axios, Safe Area Context.
• Backend – Flask, InsightFace, OpenCV, MoviePy, Google Speech + Gemini APIs.
• Storage – JSON conversation files (backend/conversations), cropped faces (backend/faces_db).

What’s Next

• Ship a secure cloud sync (currently files live on the host you deploy).
• Add inline editing so you can correct transcripts without re-uploading.
• Polish the command center with more “morning digest” style insights.

Have fun with ReCall—and let us know if you drop it into a hackathon booth. 😊