• Institution
  Upper Austria University of Applied Sciences
• Year
  2022
• Type
  Master Thesis
  Movement Analysis
  Scientific Research
• Software
  Python
• My Part
  Concept
  Research
  Implementation
  System Evaluation
  EVERYTHING

Squat Analyzer is an innovative application developed as part of my Master's thesis, titled "A Correct Form Evaluation of the Fitness Exercise Squat with Modern 3D Pose Estimation Systems." The primary objective of this project was to create an application that utilizes advanced 3D human pose estimation techniques to track and evaluate the correct form of a squat exercise in real time, providing valuable feedback on mistakes.

Through extensive research on 3D human pose estimation and the squat exercise, I identified the most common mistakes made during squats, including incorrect knee movement, compromised back posture, premature hip movement, inefficient movement path, and inadequate depth. To achieve accurate form evaluation, I evaluated several pose estimation systems using a marker-based motion tracking system. Among the options, the RGB camera-based system MeTRABs and the depth camera-based Azure Kinect camera emerged as suitable candidates.

The Squat Analyzer application was developed using Python. It is independent from any specific pose estimation system, which enables seamless integration with various systems, including both MeTRABs and Azure Kinect. To validate its performance and effectiveness, I conducted thorough testing with 30 participants, each performing 10 squats using both camera systems.

Comparing the results of mistake detection with expert assessments, Squat Analyzer proved highly effective in identifying key form deviations, including left knee path, depth, movement path, and premature hip elevation. In order to showcase the application's capabilities and its real-time mistake detection feature, I have included a slideshow on this website portfolio. The slideshow demonstrates examples of both good and bad squat forms, with clear explanations of the identified mistakes located in the top left corner.

Last Project

Next Project