Workout Results
Client Management
Welcome Screen
Start Screen
Demo

MoveMetric

Your form. Your progress. Your AI coach.

MoveMetric turns an ordinary webcam into a real-time fitness coach. It watches your movement, counts repetitions, detects form problems, creates personalized workouts, and helps you understand how you are improving.

Inspiration

Working out alone can be frustrating.

You may know what an exercise is supposed to look like, but it is difficult to know what your own body is doing while you move. A workout video can demonstrate a squat, but it cannot tell you that your knees are collapsing inward. A workout plan can tell you to complete ten push-ups, but it cannot see that your hips are dropping halfway through the set.

Personal trainers provide that feedback, but regular sessions can be too expensive or inconvenient for many students, beginners, and families.

We created MoveMetric because we believe useful fitness feedback should not depend on how much money someone has or whether another person is available to watch them train.

Our goal was simple:

Turn a device people already own into a coach that can watch every repetition.

What it does

MoveMetric creates a complete training cycle:

Check in → Receive a workout → Train with live feedback → Review your results → Improve over time

Real-time form analysis

The user chooses an exercise and positions themselves in front of their webcam. MoveMetric then:

Detects body landmarks
Draws a live skeleton over the camera
Calculates important joint angles
Recognizes each stage of the movement
Counts completed repetitions
Measures range of motion
Detects exercise-specific form problems
Gives immediate corrective feedback
Compares the left and right sides of the body
Saves performance data for later review

Instead of discovering a problem after the workout, the user can correct their very next repetition.

For example, MoveMetric can detect:

A squat that is too shallow
Knees collapsing inward
Hips sagging during a push-up
Excessive backward lean during a shoulder press
One arm rising higher than the other
Momentum being used during a bicep curl
Incomplete hip extension during a glute bridge
Poor alignment during a plank

Sixteen supported exercises

MoveMetric currently includes specialized tracking for:

Squats
Push-ups
Bicep curls
Shoulder presses
Lateral raises
Lunges
Romanian deadlifts
Glute bridges and hip thrusts
Calf raises
Planks
Jumping jacks
Mountain climbers
High knees
Sit-ups
Wall sits
Side planks

Most exercises use repetition tracking. Planks, wall sits, and side planks use form-aware timers.

More than a rep counter

Our first repetition counter used a simple joint-angle threshold. During testing, we noticed that slow movements could be counted more than once and incomplete repetitions could sometimes be accepted.

We replaced this with exercise-specific movement states.

For a squat, the system tracks whether the user is standing, lowering, reaching the bottom position, and returning to the top. A repetition is only counted after the complete movement sequence is finished.

We also use:

Separate entry and exit thresholds
Cooldown frames
Hysteresis zones
Minimum range-of-motion requirements
Landmark-confidence filtering

This made the repetition counting much more stable.

Personalized workouts

Before starting, the user completes a short readiness check-in.

MoveMetric considers:

Energy level
Hours of sleep
Muscle soreness
Available workout time
Fitness level
Fitness goal
Available equipment
Home or gym access
Recent workout history

The app then creates a workout that fits the user's current situation.

For example, if someone reports poor sleep, low energy, and major leg soreness, MoveMetric can lower the workout volume and avoid heavily training the same muscles again.

Post-workout results

At the end of a session, MoveMetric creates a clear summary containing:

Repetitions completed
Workout duration
Form score
Detected form problems
Movement balance
Possible fatigue events
Estimated calories burned
Muscle groups trained

The user also receives feedback explaining what went well, what needs the most improvement, and what to focus on during the next workout.

Progress tracking

MoveMetric saves workout history so progress is not measured only through body weight.

The dashboard can show:

Workout frequency
Total repetitions
Exercise duration
Average form score
Estimated calories
Muscle groups trained
Body-weight history
Left-versus-right balance
Common form mistakes
Recent fatigue patterns

A user may not see a major physical transformation immediately, but they may already be moving with better control and making fewer mistakes. MoveMetric makes that improvement visible.

Nutrition guidance

MoveMetric also generates estimated nutrition targets based on the user's body information, activity level, and fitness goal.

The nutrition section includes:

Estimated daily calories
Protein targets
Carbohydrate targets
Fat targets
Hydration guidance
Example meals
Simple explanations

These recommendations are educational estimates and are not intended to replace professional medical or dietary advice.

Human-coach connection

We did not want MoveMetric to separate users from real coaches.

Computer vision can watch every repetition, but a human coach can understand motivation, previous injuries, personal limitations, and long-term goals.

Users can request to connect with a coach. Once accepted, the coach can:

Review workout results
View form scores
Identify repeated mistakes
Examine progress over time
Assign workout plans
Leave personalized notes

This allows people to train independently while still receiving human guidance when needed.

How we built it

Frontend

MoveMetric was built using:

React 18
Vite
TypeScript
React Router
TanStack Query
Tailwind CSS
shadcn/ui
Radix UI
Framer Motion
Recharts

Computer vision

We used MediaPipe Tasks Vision and PoseLandmarker to detect body landmarks through the webcam.

For each camera frame, MoveMetric:

Detects the user's body landmarks
Smooths the landmark positions
Selects the joints needed for the exercise
Calculates joint angles
Determines the current movement phase
Updates the repetition count or hold timer
Checks for form problems
Updates the live feedback

Pose processing happens inside the browser, so the application does not have to continuously upload camera frames to a remote server.

Joint-angle calculation

MoveMetric calculates an angle using three body landmarks:

function calculateAngle(a, b, c) {
  const radians =
    Math.atan2(c.y - b.y, c.x - b.x) -
    Math.atan2(a.y - b.y, a.x - b.x);

  let angle = Math.abs((radians * 180) / Math.PI);

  if (angle > 180) {
    angle = 360 - angle;
  }

  return angle;
}