Authors:

Ziang Wang

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

A significant proportion of musicians will sustain musculoskeletal injuries in their career, with suboptimal posture frequently cited as a primary contributing factor. To reduce the risk of injuries due to poor posture for practicing cellists, we introduce Cello Evaluator, a system designed to identify key errors in postural form, including the supination of the wrist, the height of the elbow and shoulder, and the placement of the bow relative to the strings of the cello. Our approach uses Google Mediapipe for tracking hand and shoulder locations of the user and YOLOv11 for tracking the bow and strings. The coordinate points generated by these models are then used for classification: 1) two deep neural networks assess the correctness of the wrist and elbow posture 2) linear algebra techniques are applied to evaluate shoulder height, bow positioning, and bow angle. Using these features, we present an application able to identify and correct cellist posture in both real time and pre-recorded videos, thereby reducing the risk of injury for practicing cellists using our system. Future improvements would include implementing the models into mobile for improved privacy and optimizing current models for faster and more accurate inference. Keywords: Computer Vision; Machine Learning; Music

Source:

Purdue University / 2025

Topics:

No topics listed

Co-authors:

Ziang Wang