Hair Follicle Classification and Hair Loss Severity Estimation Using Mask R-CNN

The study presents an intelligent system using Mask R-CNN, a deep learning algorithm, to detect and measure hair loss severity. The system classifies hair follicles into healthy, normal, and severe categories, based on features like the number of hair follicles, hair thickness, and the number of hairs in each follicle. The system was trained on 600 images from 10 men, achieving 82% to 87% accuracy in the training dataset and 76% to 86% in the test dataset. It also calculates a hair loss severity index and visualizes hair loss using a heatmap. The study acknowledges limitations, such as difficulty in detecting all hair follicles in 2D images and the dataset's limited generalizability, as it was collected from Asian men with black hair. Future work will focus on improving detection performance and validating the method on different hair types, colors, and skin tones.