Smart Fitness Tracker using OpenCV

Overview

This tracker uses computer vision and pose detection (via MediaPipe) to evaluate common workouts: curls, push-ups, and squats.

How Detection Works

1. Bicep Curls

• Uses the angle between shoulder → elbow → wrist on the right arm.
• A rep is counted when:
  • Angle decreases below 40° (arm up)
  • Then increases past 160° (arm down)
• Tips are triggered if the elbow swings or angle exceeds natural bounds.

2. Push-ups

• Uses left shoulder → elbow → wrist angle.
• A rep is counted when:
  • Angle drops below 90° (going down)
  • Then exceeds 150° (coming up)
• Tips correct shallow reps or collapsing posture.

3. Squats

• Uses hip → knee → ankle angle on the right side.
• A rep is counted when:
  • Angle drops below 90° (squat down)
  • Then exceeds 160° (stand up)
• Tips correct depth and knee alignment.

Outputs

• CSV logs with time, mode, rep count, and bad form count.
• Tips CSV for feedback history.
• Recorded video (AVI) of the session.

More Info:

If you would like to see how the app is setup, how to use it, and various examples of me using the app, check out the informational video here: