FormFit AI

FormFit AI is an AI-powered fitness app built with Flutter. It uses on-device pose detection to watch you exercise in real time, corrects your form, and counts only the reps you do correctly.

Train. Compete. No dishonest reps.

---

✨ Features

• Real-time pose detection — Google ML Kit tracks your body through the camera, no internet required during the workout.
• AI-powered rep counting — push-ups are counted only when the form is actually correct: full depth reached, body kept straight, arms locked out at the top.
• Live form coaching — on-screen cues like "Go lower", "Lift your hips" and "Good form" guide you while you train.
• Side-view and front-view support — works whether the phone is propped on the floor beside you or facing you.
• Session summary — reps, duration and a form-accuracy score (blended from rep quality and time spent in good form) after every workout.
• History — Daily / Weekly / Monthly views with a tappable bar chart, rest-day tracking and per-day breakdowns.
• Daily goal ring — set a personal rep target and track progress on the Home screen.
• Leaderboard — This Week and All Time boards, with an animated spotlight on the top 3 (gold / silver / bronze shine and glow).
• Profile — Google or Guest sign-in, editable display name, custom avatar colour, and a profile photo you can view full-screen or change from the gallery.
• Appearance — System / Light / Dark theme, saved on the device.
• Resilient saving — if you're offline, your workout is queued and uploaded automatically once you're back online; if a save genuinely fails, you get a clear retry prompt instead of silently losing your workout.
• Workout session UX — 3‑2‑1 countdown before counting starts, pause/resume mid-session, and the screen stays awake for the whole workout.

---

🛠 Tech stack

Layer	Technology
App framework	Flutter (Dart)
State management	Riverpod
Navigation	go_router
Auth	Firebase Authentication (Google Sign‑In + Anonymous/Guest)
Database	Cloud Firestore
Pose detection	google_mlkit_pose_detection
Camera	camera
Local storage	shared_preferences

---

📸 Screenshots

Home	Profile	Leaderboard (This Week)	Leaderboard (All Time)

---

🚀 Getting started

Prerequisites

• Flutter SDK (stable channel)
• A physical Android or iOS device — pose detection needs a real camera and does not work well on an emulator
• A Firebase project with Authentication (Google + Anonymous providers) and Cloud Firestore enabled

Setup

# Clone the repo
git clone https://github.com/AnasXCode/formfit-ai.git
cd formfit-ai

# Install dependencies
flutter pub get

# Run on a connected device
flutter run

Firebase configuration

This project uses FlutterFire for Firebase setup. lib/firebase_options.dart is already generated for this project's Firebase instance. To connect your own Firebase project instead:

dart pub global activate flutterfire_cli
flutterfire configure

Firestore security rules

Each user should only be able to read the leaderboard and write to their own document:

rules_version = '2';
service cloud.firestore {
  match /databases/{database}/documents {
    match /users/{uid} {
      allow read: if request.auth != null;
      allow create, update: if request.auth != null && request.auth.uid == uid;

      match /sessions/{sessionId} {
        allow read, create: if request.auth != null && request.auth.uid == uid;
      }
    }
  }
}

⚠️ Treat these as a starting point, not production-ready rules — review them before shipping.

---

📂 Project structure

lib/
├── models/        # Data models (WorkoutSession, UserProfile, Exercise, ...)
├── pose/          # Pose-detection logic: angle math, rep counting, skeleton overlay
├── providers/      # Riverpod providers (auth, sessions, leaderboard, theme, goals)
├── router/         # go_router route definitions
├── screens/        # Top-level app screens
├── theme/          # Colours, text styles, ThemeData
├── widgets/         # Reusable UI components
├── firebase_options.dart
└── main.dart

---

🧠 How rep counting works

Push-up form is evaluated from the joint positions ML Kit returns each frame:

1. Elbow angle (shoulder → elbow → wrist) tracks how low you go. A rep only counts once the elbow bends past a set depth threshold.
2. Body-line angle (shoulder → hip → ankle) checks that the body stays straight — sagging or piking hips rejects the rep.
3. A short frame-stability window filters out single noisy detections so one bad frame can't falsely count — or reject — a rep.

The same frames feed a form score: part rep accuracy (good reps ÷ attempted reps), part time spent in good form while actively exercising.

---

🗺 Roadmap

• Voice feedback ("go lower", rep counts) so you don't need to watch the screen
• Additional exercises (squats, sit-ups, lunges, plank)
• Auto-calibration to each user's own range of motion
• Friends & challenges
• Workout reminders/notifications
• Achievements and badges

---

🤝 Contributing

Issues and pull requests are welcome. For larger changes, please open an issue first to discuss what you'd like to change.

📄 License

No license has been chosen yet for this project. All rights reserved by the author until one is added.

👤 Author

Anas Ahmed — @AnasXCode