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

A hairstyle recommendation system that would recommend hairstyles according to facial shapes or other properties could be useful for barbers and their customers alike. However, currently, there are no datasets with attributes necessary for this task. Therefore, a recent paper introduces a new large-scale dataset comprising more than 200 000 facial images with the corresponding hairstyles and attributes like face shape, nose length, or pupillary distance.

Image credit: pxhere.com, CC0 Public Domain

In the process of feature extraction, facial landmark detection, convolutional neural networks, and spatial transformer networks are used. As a validation, a hairstyle recommendation system based on the Random Forests algorithm is proposed. It predicts the hairstyle from facial features and lets users also try on a hairstyle. These applications confirm the robustness and usability of the suggested dataset.

In this paper, we present a new large-scale dataset for hairstyle recommendation, CelebHair, based on the celebrity facial attributes dataset, CelebA. Our dataset inherited the majority of facial images along with some beauty-related facial attributes from CelebA. Additionally, we employed facial landmark detection techniques to extract extra features such as nose length and pupillary distance, and deep convolutional neural networks for face shape and hairstyle classification. Empirical comparison has demonstrated the superiority of our dataset to other existing hairstyle-related datasets regarding variety, veracity, and volume. Analysis and experiments have been conducted on the dataset in order to evaluate its robustness and usability.

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

Source

Subscribe
Notify of
0 Comments
Inline Feedbacks
View all comments
0
Would love your thoughts, please comment.x
()
x