[4I3-GS-2] Machine learning: Image recognition and video analysis

[4I3-GS-2-04] Identification Method of Dance Movie Genres using OpenPose and Autoencoder

〇Ryuku Nobusue¹, Masanori Akiyoshi¹, Jun Ichikawa¹ (1. Kanagawa University)

Keywords:Deep Learning, Autoencoder, Classification

Since "Dance" becomes compulsory subjects, teachers who are not
trained before and have anxiety on instruction are apt to join lesson
classes or training seminar, which leads to take time and money. To
overcome this situation, there exists strong needs to learn knowledge
and skill as to "Dance" for themselves through watching free-of-charge
internet dance movies. First of all it is necessary to discriminate
dance genres for the sake of watching such movies from explicit move-
conscious viewpoints. This paper addresses an identification method
of dance movie genres using OpenPose and Autoencoder.