Machine Learning Enabled 3D Body Measurement Estimation Using Hybrid Feature Selection and Bayesian Search

The 3D body scan technology has recently innovated the way of measuring human bodies and generated a large volume measurements. However, one inherent issue that plagues use resultant database is missing data usually caused by using automatic extractions from scans. Tedious extra efforts have to be made manually fill for various applications. To tackle this problem, paper proposes machine learning (ML)-based approach measurement estimation while considering (feature) importance. proposed selects most critical features reduce algorithm input improve ML method performance. In addition, Bayesian search further used in fine-tuning hyperparameters minimize mean square error. Two distinct methods, i.e., Random Forest XGBoost, are tested on real-world dataset contains scans 212 participants Kansas-Missouri area United States. results show effectiveness methods with roughly 3% Mean Absolute Percentage Errors estimating data. two hybrid feature selection Baysian comprehensively compared. comparative suggest performs better than XGBoost counterpart filling