EQUINE BACK-SURFACE MODELING AND SHAPE CLASSIFICATION USING ML-TECHNIQUES

Abstract

The current state of saddle manufacturing faces several challenges. The main concern is the lack of a universally accepted horse-back measurement system that makes the whole manufacturing process time consuming and labor intensive. The aim of this research is to combine parametric modeling with Machine Learning techniques for shape classification to develop a software tool which will make saddle design and its manufacturing process more efficient. To develop an appropriate neural network, the raw dataset containing horse-back measurements is required to firstly develop parametric models with appropriate design variables and ultimately corresponding saddle surfaces. Using geometric properties of the surfaces, a number of features will be extracted in order to develop and train the corresponding classification algorithm. Several categories will be identified based on surface shape characteristics and representative surfaces for each category will be produced. The results obtained during the course of this research will be used to create manufacturable saddle models that fit performance standards and maintain the well-being of horses.