CelebHair: A New Large-Scale Dataset for Hairstyle Recommendation based on CelebA

Victoria D. Doty

A hairstyle recommendation program that would advocate hairstyles in accordance to facial designs or other houses could be handy for barbers and their customers alike. On the other hand, at this time, there are no datasets with attributes needed for this undertaking. Thus, a new paper introduces a new massive-scale dataset comprising a lot more than 200 000 facial illustrations or photos with the corresponding hairstyles and attributes like experience condition, nose duration, or pupillary distance.

Taking photo with a smartphone. Image credit: pxhere.com, CC0 Public Domain

Image credit history: pxhere.com, CC0 Public Area

In the course of action of attribute extraction, facial landmark detection, convolutional neural networks, and spatial transformer networks are applied. As a validation, a hairstyle recommendation program primarily based on the Random Forests algorithm is proposed. It predicts the hairstyle from facial characteristics and allows customers also try out on a hairstyle. These programs affirm the robustness and usability of the recommended dataset.

In this paper, we present a new massive-scale dataset for hairstyle recommendation, CelebHair, primarily based on the celebrity facial attributes dataset, CelebA. Our dataset inherited the the vast majority of facial illustrations or photos alongside with some attractiveness-linked facial attributes from CelebA. Also, we used facial landmark detection techniques to extract excess characteristics these as nose duration and pupillary distance, and deep convolutional neural networks for experience condition and hairstyle classification. Empirical comparison has shown the superiority of our dataset to other present hairstyle-linked datasets concerning wide range, veracity, and quantity. Investigation and experiments have been conducted on the dataset in order to consider its robustness and usability.

Investigation paper: Chen, Y., Zhang, Y., Huang, Z., Luo, Z., and Chen, J., “CelebHair: A New Massive-Scale Dataset for Hairstyle Advice primarily based on CelebA”, 2021. Link: https://arxiv.org/abs/2104.06885


Next Post

Designing better antibody drugs with artificial intelligence

Machine discovering solutions support to optimise the advancement of antibody prescription drugs. This sales opportunities to energetic substances with enhanced properties, also with regard to tolerability in the entire body. Machine discovering solutions support to optimise the advancement of antibody prescription drugs. This sales opportunities to energetic substances with enhanced […]

Subscribe US Now