A study of feature extraction methods for clusters in image classification using deep metric learning

－Visualization of features using factor information common to clusters－

Abstract

In recent years, the running population has been increasing, and demand for coaching systems for amateur runners is expected to increase. Since each individual has a different goal and body shape, it is necessary to provide more personalized coaching, which requires quantitative evaluation of the characteristics of individual movements. As a first step, this paper examines a method for extracting and showing characteristic shapes common to clusters from still image data of common footwear using deep metric learning. In our experiments, we were able to show the common characteristic shapes for the entire cluster based on the common features in the set according to the distance from the cluster center. The application of the experimental results to running behavior is discussed.