Abstract:

Artificial intelligence (AI) continues to transform the beauty and cosmetic industry by enabling customized recommendations and automated analysis. This paper presents an eye image categorization and makeup recommendation system built using deep learning based on the YOLOv8 model. A private dataset of male and female eye images was preprocessed and split into training and validation sets for equitable learning. The YOLOv8 classifier was trained to distinguish between categories of eyes with precise consistency, as context-aware makeup recommendations’ foundation. In addition to categorization, the system incorporates a built-in rule-based recommendation component that provides two forms of personalization: generalized makeup styling recommendations based on gender-specific eye features, and eyeshadow color suggestions as a function of user-provided skin tone indicators (fair, medium, or deep). Two-stage use of computer vision and cosmetics expertise generates more significant and user-friendly suggestions compared to conventional static approaches. The proposed system has application in virtual beauty assistants, web shops, and mobile applications, offering scalable and responsive solutions for enhanced customer interaction. By integrating AI-based classification and cosmetic domain expertise, this study underscores the growing presence of intelligent systems in the work of personalized beauty technology.