سامانه همایش‌ها TSTA | ثبت‌نام و ارسال مقاله کنفرانس‌ها و کنگره‌های ملی

صفحه اصلی / سی و دومین کنفرانس ملی و دهمین کنفرانس بین المللی مهندسی زیست پزشکی ایران

Automated Kinematic Analysis of Barbell Curl Using Custom IMU and Deep Learning Techniques

نویسندگان :

Mohammad Khalfe Nilsaz¹ Elham Shirzad² Ali Fahim³

1- دانشگاه تهران دانشکده علوم مهندسی 2- دانشگاه تهران 3- دانشگاه تهران دانشکده علوم مهندسی

کلمات کلیدی :

IMU،Machine Learning،Convolutional Neural Networks،Low Back Injury،Barbell Curl

چکیده :

In this article, we present a method for monitoring barbell motion during biceps curl exercises to assess movement accuracy and identify excessive stress on the lumbar spine in order to reduce injury risk. A custom IMU sensor was designed, validated, and mounted on the barbell using dedicated clamps. Data were collected from 53 athletes with diverse sporting backgrounds under three conditions: light execution, moderate intensity, and maximum effort. Using a Convolutional Neural Network (CNN), the system achieved an overall accuracy of 92.2%. Specifically, correct repetitions were detected with a precision of 96.6% and a recall of 94.7%, while incorrect repetitions reached a precision of 60% and a recall of 70.6%. These results highlight the potential of sensor-based monitoring as a practical alternative to coach-supervised methods for evaluating exercise performance, while also indicating areas where future improvements are needed. The primary objective of this study was to detect sudden impulses along the anterior–posterior axis, as such loads are known to contribute to intervertebral disc injuries.