Skip to content
asifulshiamPublic

About

AI-powered photo meal tracker — Flutter + FastAPI + Gemini Vision

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

5 Commits

Folders and files

Repository files navigation

MealSnap — AI Nutrition Tracker

Snap a photo of your meal, get instant AI-powered nutrition analysis, and track your diet with a personal AI assistant that knows your pantry.

Live Web App: https://mealsnap-327.web.app  |  API Docs: https://mealsnap-8bgs.onrender.com/docs

Note: Backend runs on Render free tier — first request after idle takes ~30s to wake up.


Screenshots

Sign In Create Account Goal Setup (TDEE) Dashboard
Camera / Log Meal AI Analyzing Detection Results Quantity Slider
AI Chat Assistant Pantry Add Pantry Item Recipe Suggestions
Meal History Empty Dashboard Group Chat Rooms Real-time Chat
Create Room + Invite Friendly Error

Features

Core

  • Photo-based meal logging — take a photo or pick from gallery; Gemini Vision identifies every food item with confidence scores (High / Medium / Low)
  • 60 FPS quantity slider — drag to adjust grams per food item; calories and macros recalculate in real time as you drag
  • USDA nutrition data — calorie and macro values sourced from USDA FoodData Central, cached to avoid duplicate lookups
  • Daily nutrition dashboard — calorie ring, macro breakdown (protein / carbs / fat), and a live list of today's meals
  • Meal history — browse all past meals filtered by today / this week / this month; tap any meal for full item detail

AI

  • AI Nutrition Assistant — conversational chat powered by Gemini with full function-calling: ask questions, log meals, manage pantry, and get a daily summary — all through natural language
  • Context-aware responses — the assistant knows your recent meals and pantry contents before it replies
  • Time-context injection — current time is sent with every photo analysis so Gemini can distinguish breakfast from dinner

Pantry & Recipes

  • Pantry management — add ingredients with quantity, unit, category, and expiration date; expiring-soon alerts surface automatically
  • AI recipe suggestions — tap Recipes to get Gemini-generated meal ideas based on exactly what you have in your pantry

Social

  • Real-time group chat — Firestore-backed multi-room chat; create a room, invite friends by email, messages appear instantly via StreamBuilder

UX

  • Latency masking — rotating funny messages + progressive step indicators make the 2–4s AI analysis feel fast
  • TDEE onboarding — Mifflin-St Jeor formula calculates personalised daily calorie and macro targets after registration
  • Friendly error messages — no raw JSON dumps; every error (no food detected, AI timeout, network failure) gets a human-readable dialog
  • Dark theme — "Neural Night" dark theme across all screens

Tech Stack

Layer Technology
Frontend Flutter 3.38 (Web + Android)
Backend FastAPI + Python 3.12
Database PostgreSQL 15 (SQLAlchemy ORM)
AI — Vision Google Gemini 2.5 Flash (photo analysis)
AI — Chat Google Gemini 2.5 Flash (function-calling chat)
Nutrition Data USDA FoodData Central API
Auth JWT (HS256 + PBKDF2, no external deps)
Real-time Chat Firebase Firestore + Anonymous Auth
Frontend Hosting Firebase Hosting
Backend Hosting Render (free tier)

Architecture

┌─────────────────────────────────────────────────────┐
│                Flutter Frontend                     │
│  Web (Firebase Hosting) + Android APK               │
│                                                     │
│  screens/   widgets/   services/   models/          │
│     │           │           │                       │
│     └───────────┴───────────┘                       │
│              api_service.dart  ←── JWT auth         │
│              firebase_chat_service.dart ←── Firestore│
└────────────────────┬────────────────────────────────┘
                     │ HTTPS
┌────────────────────▼────────────────────────────────┐
│              FastAPI Backend (Render)               │
│                                                     │
│  /api/auth        JWT register / login / me         │
│  /api/meals       analyze (Gemini Vision + USDA)    │
│  /api/chat        Gemini function-calling chat      │
│  /api/pantry      CRUD + expiring-soon              │
│  /api/group-chat  legacy HTTP group messages        │
│                                                     │
│  services/gemini.py       ← Gemini Vision API       │
│  services/gemini_chat.py  ← Gemini Chat + tools     │
│  services/nutrition.py    ← USDA lookup + cache     │
│  middleware/rate_limit.py ← token bucket (burst 3)  │
└────────────┬───────────────────┬────────────────────┘
             │                   │
    ┌────────▼──────┐   ┌────────▼──────────┐
    │  PostgreSQL   │   │  Firebase          │
    │  (Render)     │   │  Firestore + Auth  │
    │  8 tables     │   │  (group chat)      │
    └───────────────┘   └────────────────────┘

Local Setup

Prerequisites

1. Clone & configure

git clone https://github.com/asifulshiam/mealsnap.git
cd mealsnap

Create backend/.env:

GEMINI_API_KEY=your_gemini_key_here
USDA_API_KEY=your_usda_key_here
DATABASE_URL=postgresql://localhost/mealsnap_db
JWT_SECRET=your_random_secret_here

2. Backend

# Create conda environment
conda create -n mealsnap python=3.12 -y
conda activate mealsnap
pip install -r backend/requirements.txt

# Create the database
createdb mealsnap_db

# Start the server
cd backend
uvicorn app.main:app --reload
# API docs → http://localhost:8000/docs

3. Frontend

cd frontend
flutter pub get
flutter run -d chrome

The app defaults to http://localhost:8000 as the backend URL for local dev.

4. Verify


Project Structure

mealsnap/
├── backend/
│   └── app/
│       ├── main.py              # FastAPI app, CORS, router registration
│       ├── database.py          # SQLAlchemy engine + table creation
│       ├── models/              # ORM models (user, meal, food, pantry, chat)
│       ├── schemas/             # Pydantic request/response schemas
│       ├── routers/             # auth, meals, chat, pantry, group_chat
│       ├── services/
│       │   ├── gemini.py        # Gemini Vision — photo → food list
│       │   ├── gemini_chat.py   # Gemini Chat — function-calling loop
│       │   └── nutrition.py     # USDA lookup with in-memory cache
│       ├── middleware/
│       │   └── rate_limit.py    # Token bucket (burst 3, 1/4s refill)
│       └── tests/               # pytest suite (36 tests, real DB)
├── frontend/
│   └── lib/
│       ├── main.dart
│       ├── screens/             # 18 screen files
│       ├── widgets/
│       │   └── smooth_quantity_slider.dart  # 60 FPS killer feature
│       ├── services/
│       │   ├── api_service.dart             # HTTP client + JWT
│       │   └── firebase_chat_service.dart   # Firestore group chat
│       └── theme/app_theme.dart             # "Neural Night" dark theme
└── docs/
    ├── screenshots/             # All app screenshots
    └── MEALSNAP_CONTEXT.md      # Full technical reference

API Overview

Method Endpoint Description
POST /api/auth/register Create account → JWT
POST /api/auth/login Login → JWT
GET /api/auth/me Current user from token
POST /api/meals/analyze Photo → Gemini Vision → USDA → save
GET /api/meals List user's meals
GET /api/meals/summary/daily Daily totals + meal list
POST /api/chat Message → Gemini function-calling → response
GET /api/pantry List pantry items
POST /api/pantry Add pantry item
GET /api/pantry/expiring?days=3 Items expiring within N days

Full interactive docs at /docs when running locally.


Deployment

Service Platform URL
Web frontend Firebase Hosting (Spark — free) https://mealsnap-327.web.app
Backend API Render (free tier) https://mealsnap-8bgs.onrender.com
Database Render PostgreSQL (free tier) internal to backend
Group chat Firebase Firestore mealsnap-327 project

To redeploy the backend, push to feature/frontend-ui-complete — Render auto-deploys on every commit.

To redeploy the frontend:

cd frontend
nvm use --lts
flutter build web --release --dart-define=API_URL=https://mealsnap-8bgs.onrender.com
firebase deploy --only hosting --project mealsnap-327

About

AI-powered photo meal tracker — Flutter + FastAPI + Gemini Vision

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages