Deep-Learning-Based Hair Damage Diagnosis Method Applying Scanning Electron Microscopy Images.

Area of Science:

• Dermatology and Cosmetology
• Artificial Intelligence in Healthcare
• Materials Science

Background:

• Assessing hair damage is crucial in beauty services and medicine due to frequent chemical and physical treatments.
• Current methods like visual inspection, microscopy, and physical/chemical tests are time-consuming, complex, and inconvenient.
• There is a need for a rapid, accurate, and efficient method for hair damage diagnosis.

Purpose of the Study:

• To develop and validate a deep learning-based method for precise hair damage degree assessment.
• To establish a new standard for hair quality diagnosis using advanced imaging and AI.
• To overcome the limitations of existing hair damage evaluation techniques.

Main Methods:

• Utilized scanning electron microscopy (SEM) to capture detailed images of hair samples.
• Developed a deep learning model trained on a custom dataset of hair images.
• Compared the proposed deep learning model's performance against other lightweight networks.

Main Results:

• The proposed deep learning method achieved a high accuracy rate of 94.8% in identifying and judging hair damage.
• Demonstrated superior performance compared to existing lightweight network experimental results.
• Successfully applied the method for hair quality diagnosis on the created dataset.

Conclusions:

• The combination of SEM imaging and deep learning offers a highly accurate and efficient solution for hair damage assessment.
• This AI-driven approach provides a reliable tool for hair quality diagnosis in cosmetic and medical fields.
• The developed method significantly advances the field by offering a faster and more precise alternative to traditional evaluation techniques.