Gender classification based on hair style?

[red010182]

This is not a bug issue.

Just curious about that is it possible to consider the hair style as a factor for gender classification?

Because in real world, most (not 100%) women are with longer hair than men. Seeing one person with hair long below shoulder, pretty much I'll guess his/her gender a higher probability to be a female.

Thanks.

[GilLevi]

Hi @red010182 ,

Thank you for your interest in our work. That's an interesting question. Indeed, hair style sounds like an important feature to consider.

I think that if you would not crop the images too aggressively so that after cropping the image would have the face as well as the hair, then the model trained on those images would implicitly take into account the hair style.

In fact, it's a nice experiment. You can train a network like that and then use the method of Zeiler and Fergus[1] to visualize what are the most important features in the image that the network have leared and see if indeed the hair style turns out to be an important feature.

Best, Gil

[1] Zeiler, Matthew D., and Rob Fergus. "Visualizing and understanding convolutional networks." European conference on computer vision. Springer, Cham, 2014.‏