Smart Fashion Advisor: AI-Powered Skin Tone Analysis and Personalized Outfit Recommendation Using Computer Vision

Abstract

 The rapid growth of artificial intelligence in lifestyle applications has significantly influenced the fashion industry by enabling personalized and data-driven recommendations. This paper presents a Smart Fashion Advisor, an intelligent system designed to assist users in selecting suitable outfits based on their skin tone, preferences, and contextual factors such as occasion and season. The system integrates computer vision techniques with data-driven filtering methods to provide real-time fashion suggestions. By utilizing facial image analysis, the application automatically detects the user’s skin tone and maps it to a predefined set of color palettes, thereby ensuring recommendations that enhance visual appeal and personal style.

 The proposed system employs MediaPipe-based facial landmark detection with 468 facial points and OpenCV image processing to extract relevant facial regions for accurate skin tone classification. Unlike purely deep learning-based recommendation systems, this approach combines rule-based color theory with a large-scale fashion dataset of over 44,000 items to generate meaningful outfit suggestions. The dataset is processed using efficient Pandas-based filtering techniques based on attributes such as gender, category, color, and usage. Additionally, the system incorporates a web-based interface built using Flask, enabling users to upload images or use a live camera feed to receive instant recommendations. An optional AI-powered chatbot using the Google Gemini API further enhances user interaction by providing styling advice and answering fashion-related queries.

 By automating the outfit selection process, the Smart Fashion Advisor reduces decision-making effort and improves personalization in everyday fashion choices. The system demonstrates the effectiveness of combining computer vision, domain knowledge, and data filtering in building practical recommendation systems. It can be applied in various domains such as e-commerce platforms, virtual styling assistants, and personal wardrobe management tools. Future enhancements may include integrating deep learning-based recommendation models, real-time trend analysis, and augmented reality features to further improve user experience and system intelligence.

 KEYWORDS: Computer Vision, Skin Tone Detection, Fashion Recommendation, MediaPipe, OpenCV, Color Theory, Flask, Machine Learning, Personalized Styling, Outfit Recommendation, Image Processing, Chatbot, Gemini API.