ScalpEye: A Deep Learning-Based Scalp Hair Inspection and Diagnosis System for Scalp Health

Abstract

Many people suffer from scalp hair problems such as dandruff, folliculitis, hair loss, and oily hair due to poor daily habits, imbalanced nutritional intake, high stress, and toxic substances in their environment. To treat these scalp problems, dedicated services such as scalp hair physiotherapy have emerged in recent years. This article proposes a deep learning-based intelligent scalp inspection and diagnosis system, named ScalpEye, as an efficient inspection and diagnosis system for scalp hair physiotherapy as part of scalp healthcare. The proposed ScalpEye system consists of a portable scalp hair imaging microscope, a mobile device app, a cloud-based artificial intelligence (AI) training server, and a cloud-based management platform. The ScalpEye system can detect and diagnose four common scalp hair symptoms (dandruff, folliculitis, hair loss, and oily hair). In this study, we tested several popular object detection models and adopted a Faster R-CNN with the Inception ResNet_v2_Atrous model in the ScalpEye system for image recognition when inspecting and diagnosing scalp hair symptoms. The experimental results show that the ScalpEye system can diagnose four common scalp hair symptoms with an average precision (AP) ranging from 97.41% to 99.09%.

Keywords

• Artificial intelligence over the Internet of Things (AIoT)
• deep learning
• image processing
• image recognition
• inspection
• haircare