Ramón Guevara, Pablo
Prognosticating Physique: Machine Learning for Future Body Shape Estimations in Weight Loss
URI: http://hdl.handle.net/10045/136544
DOI: 
ISSN: 
Abstract: 
This research presents the development of a predictive model to forecast morphological changes in individuals undergoing weight loss treatment. The initiative, Tech4Diet, draws from the public health imperative to address the global obesity crisis and utilized 3D body scans and supplementary medical data to enhance adherence to treatment. An extensive review of the current literature on 3D human body model representation forms the foundation of this work, leading to the selection of the Skinned Multi-Person Linear Model (SMPL) model for encoding body scans. Long Short-Term Memory (LSTM) networks are employed to analyze these encoded datasets and predict potential body changes before the treatment concludes. The process includes a comprehensive analysis of collected data, body model representation, neural network design, model training, and evaluation. The resulting model successfully generates 3D meshes of predicted body transformations, offering a novel approach to visualizing weight loss progress. Further chapters detail the data acquisition, model design, training process, and results.
Keywords:Tech4Diet, Deep learning, SMPL, Obesidad, Modelo de cuerpo humano
info:eu-repo/semantics/bachelorThesis