Skin Condition Classification Using CNN with Personalized Recommendation Engine

Abstract

The increasing demand for personalized skincare solutions has led to the integration of artificial
intelligence in dermatological analysis. This paper presents an intelligent system for automatic
skin type detection and skin condition classification using deep learning techniques. The
proposed model processes facial images uploaded through a web-based interface and analyzes
them using advanced image processing and convolutional neural network (CNN) architectures.
The system identifies key skin attributes such as oiliness, dryness, acne, and other dermatological
conditions with improved accuracy and consistency.
Following the classification phase, a recommendation module generates personalized skincare
suggestions tailored to the detected skin type and condition. This module enhances user
experience by providing targeted product recommendations, thereby bridging the gap between
dermatological assessment and practical skincare solutions. The application is implemented using
a Flask-based web framework, ensuring accessibility and real-time interaction for end users.
Experimental evaluation demonstrates that the proposed system achieves reliable performance in
both classification and recommendation tasks, making it suitable for practical deployment in
digital healthcare platforms. The integration of deep learning with a recommendation engine
offers a scalable and efficient approach to personalized skincare, reducing dependency on manual
diagnosis and enabling users to make informed decisions regarding their skin health.