International Display Workshops General Incorporated Association

[AISp1-1] Real-Time Gender Classification from Gait Features Using Convolutional Neural Networks

*Kazuki Mizutani1 (1.Gifu University (Japan))

Gender Classification, Gait Silhouette, Average Gait Image, Convolutional Neural Network

I use walking silhouettes, which are not affected by wearing a mask, to classify gender. With conventional methods, only angle-dependent discriminators existed. However, by using deep learning, I have succeeded in creating a discriminator that does not depend on the angle and has the same accuracy.