Spine Posture Detection for Office Workers With Hybrid Machine Learning

Abstract

This study aims to detect bad spine posture using an al-ternative approach that doesn't rely on deep learning or excessive energy. The goal is to improve accuracy and effectiveness without disrupting workflow. A custom dataset was created, numerical inferences were made from posture values, and a hybrid approach using Light Gradient Boosting achieved a 96 % success rate.