Development of CC Fit: A Motion Analysis Fitness App
Budget: $400 – $1,000 USD
Flutter Developer Needed – Fitness App with Pose Detection (Motion Comparison Engine)
Project Overview
We are looking for a Flutter developer to build an APP for a mobile fitness application called CC Fitfor Android and iOS.
The app guides users through workout exercises and analyzes body movement during the exercise using the front camera.
Pose detection must run locally on the device using:
MoveNet Lightning – TensorFlow Lite
No video or images should be uploaded to the server.
The app should follow an offline-first approach, allowing workouts and results to be stored locally when internet is unavailable.
Budget: FROM 400 TO 1000 $
Timeline: Flexible
Tech Stack
Framework
Flutter (Dart)
Expected packages:
camera
tflite_flutter
video_player
flutter_secure_storage
Hive or SQLite
Backend:
Supabase
Pose Detection Model:
MoveNet Lightning (TensorFlow Lite)
Important Project Scope
The motion signatures for exercises will be provided by us.
Each motion signature represents the reference movement for a single repetition of an exercise.
The developer does NOT need to create motion signatures or train any AI model.
The main task is to implement a Motion Comparison Engine that compares the user's movement with the provided reference signature.
Core Features
Authentication
Simple login using:
Username
6-digit PIN
Session stored locally using flutter_secure_storage and valid for 7 days.
Dashboard
Main screen displaying:
User profile
Today’s workout
Start workout session
Points / leaderboard (basic)
Workout Program
Workout programs are downloaded from the platform server and cached locally for offline use.
Each workout contains:
Exercises
Sets
Repetitions
Instruction video
Exercise videos must be downloaded and stored locally.
Each exercise also includes a motion signature file representing the reference movement for one repetition.
Calibration Step
Before each exercise the user performs a short calibration step.
Flow:
stand straight
face the camera
arms relaxed
hold for 3 seconds
Purpose:
detect body position
calculate body scale
establish a neutral pose reference
Exercise Flow
Instruction video
↓
Calibration
↓
Countdown
↓
Camera starts
↓
Pose detection
↓
Motion comparison
↓
Repetition counting
↓
Exercise result
The user does not see the camera preview.
The camera runs only in the background for motion analysis.
Motion Comparison Engine (Main Task)
The application must implement a motion comparison system that:
Receives pose keypoints from MoveNet continuously.
Normalizes the movement using body scale.
Compares the user movement with the provided motion signature.
Detects the start and end of a repetition.
Counts repetitions.
Evaluates movement quality.
Comparison should preferably rely on:
joint angles
movement sequence
rather than raw pixel coordinates.
Motion Normalization
The system must normalize motion so the analysis remains accurate even if the user:
moves closer to the camera
moves farther from the camera
Normalization may include:
body scale calculation
normalized joint distances
angle-based comparison
Performance Goal
Pose detection must run fully on-device.
Target performance:
15–20 FPS analysis speed.
Data Storage
Workout results should be:
Saved locally first
Synced with the backend later when internet is available.
Only the following data is uploaded:
repetition_count
movement_quality
score
No video or image data should be stored or transmitted.
Preferred Experience
Flutter camera processing
TensorFlow Lite
Pose detection
Fitness / motion tracking apps
Project Overview
We are looking for a Flutter developer to build an APP for a mobile fitness application called CC Fitfor Android and iOS.
The app guides users through workout exercises and analyzes body movement during the exercise using the front camera.
Pose detection must run locally on the device using:
MoveNet Lightning – TensorFlow Lite
No video or images should be uploaded to the server.
The app should follow an offline-first approach, allowing workouts and results to be stored locally when internet is unavailable.
Budget: FROM 400 TO 1000 $
Timeline: Flexible
Tech Stack
Framework
Flutter (Dart)
Expected packages:
camera
tflite_flutter
video_player
flutter_secure_storage
Hive or SQLite
Backend:
Supabase
Pose Detection Model:
MoveNet Lightning (TensorFlow Lite)
Important Project Scope
The motion signatures for exercises will be provided by us.
Each motion signature represents the reference movement for a single repetition of an exercise.
The developer does NOT need to create motion signatures or train any AI model.
The main task is to implement a Motion Comparison Engine that compares the user's movement with the provided reference signature.
Core Features
Authentication
Simple login using:
Username
6-digit PIN
Session stored locally using flutter_secure_storage and valid for 7 days.
Dashboard
Main screen displaying:
User profile
Today’s workout
Start workout session
Points / leaderboard (basic)
Workout Program
Workout programs are downloaded from the platform server and cached locally for offline use.
Each workout contains:
Exercises
Sets
Repetitions
Instruction video
Exercise videos must be downloaded and stored locally.
Each exercise also includes a motion signature file representing the reference movement for one repetition.
Calibration Step
Before each exercise the user performs a short calibration step.
Flow:
stand straight
face the camera
arms relaxed
hold for 3 seconds
Purpose:
detect body position
calculate body scale
establish a neutral pose reference
Exercise Flow
Instruction video
↓
Calibration
↓
Countdown
↓
Camera starts
↓
Pose detection
↓
Motion comparison
↓
Repetition counting
↓
Exercise result
The user does not see the camera preview.
The camera runs only in the background for motion analysis.
Motion Comparison Engine (Main Task)
The application must implement a motion comparison system that:
Receives pose keypoints from MoveNet continuously.
Normalizes the movement using body scale.
Compares the user movement with the provided motion signature.
Detects the start and end of a repetition.
Counts repetitions.
Evaluates movement quality.
Comparison should preferably rely on:
joint angles
movement sequence
rather than raw pixel coordinates.
Motion Normalization
The system must normalize motion so the analysis remains accurate even if the user:
moves closer to the camera
moves farther from the camera
Normalization may include:
body scale calculation
normalized joint distances
angle-based comparison
Performance Goal
Pose detection must run fully on-device.
Target performance:
15–20 FPS analysis speed.
Data Storage
Workout results should be:
Saved locally first
Synced with the backend later when internet is available.
Only the following data is uploaded:
repetition_count
movement_quality
score
No video or image data should be stored or transmitted.
Preferred Experience
Flutter camera processing
TensorFlow Lite
Pose detection
Fitness / motion tracking apps