|
Instrumentation for Skin Color Classification Using RGB and NIR Sensor with Automated Product Dispensing Controlled by Load Cell Feedback (1,3,4) Universitas Negeri Jakarta Abstract The novelty of this study lies in integrating computer vision, CNN-based skin classification, AI recommendation systems, voice interaction, and smart robotic vending into a single intelligent retail platform for body skincare services. Unlike conventional retail systems that operate separately or mainly in digital environments, the proposed platform combines real-time skin analysis, automated recommendations, interactive communication, and robotic product dispensing within one system. Therefore, this study aims to develop and evaluate an AI-based personalised body skincare recommendation system capable of detecting customer skin types through computer vision and automated image analysis, while automatically dispensing skincare product samples based on recommendation outcomes. Using a Design Science Research approach, the study developed a smart robotic vending prototype integrating image acquisition, AI processing, and robotic output layers. The system utilised HD cameras, controlled lighting, CNN algorithms, recommendation engines, voice interaction, and robotic dispensing technologies to support intelligent customer interaction. The findings showed that the system successfully performed automated skin classification, generated personalised skincare recommendations, and dispensed product samples effectively in real time. In addition, the integration of AI recommendations and robotic vending improved customer engagement, personalisation, and service convenience, highlighting its strong potential for smart beauty retail applications. Keywords: Smart Retail, Computer Vision, CNN-Based Skin Classification, AI Recommendation System, Robotic Vending System Topic: Applied Technology in Physics |
| IPS 2026 Conference | Conference Management System |