XAI-based Skin Hydration Detection from Selfies with Chatbot Recommendations

Abstract

Skin hydration is considered an important parameter of skin health and wellness; however, current methods of assessing and evaluating skin dehydration are based on various equipment and tools related to dermatology and cosmetics, which are costly and not easily accessible to everyone. The current paper introduces an intelligent system for selfie-based facial skin hydration/dehydration detection using AI. The system is integrated with an AI-based chatbot related to skin wellness. The system is based on an intelligent approach using which users can easily evaluate and check their facial skin hydration and dehydration status using a simple selfie image uploaded through a web interface. The system is based on deep learning and utilizes MobileNetV2 to classify facial skin as hydrated or dehydrated. The system is based on explainable AI, which helps users understand facial regions related to dehydration. The system is implemented using TensorFlow and Flask frameworks. Once the prediction stage is done, an AI-assisted chatbot offers personal skin wellness advice through its interaction with the user and offers advice based on the detected conditions and user feedback. This chatbot is capable of answering questions regarding facial skin wellness only. The proposed system offers an efficient, non-invasive, and user-friendly method for the real-time assessment of the user’s skin conditions. This system indicates the possibility of using computer vision for skin health monitoring and AI for user convenience.