A Comprehensive Survey on Vitiligo Detection Using Deep Learning

¹ Sharnbasva University Kalaburagi, Karnataka, India.
² Sharnbasva University Kalaburagi, Karnataka, India.

^* Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it.

Abstract

Vitiligo is a skin condition that produces pale patches when pigment-producing cells are lost. Because visual diagnosis can vary between clinicians and take significant time, automated approaches for detecting and outlining lesions have become important. This survey reviews both public and private datasets such as Vit2019 and collections from UC Davis Medical Centre examining their sizes, annotation styles, diversity of skin tones, and ease of access. We compare traditional approaches like support vector machines and random forests with modern convolutional neural networks and segmentation architectures such as VGG, ResNet, MobileNet, and U-Net. Major obstacles are the small number of large, open datasets, poor representation of darker skin types, and inconsistent image capture conditions. The paper summarizes common pre-processing and evaluation practices, identifies gaps in current work, and suggests directions for developing more reliable and broadly applicable vitiligo detection systems.