Scholarlink

With the rise of e-commerce, demand for personalized fashion recommendation systems has increased. This study presents a novel approach that integrates occasion awareness, user-defined color preferences, and outfit compatibility testing using advanced machine learning and computer vision. The system employs K-means clustering for color extraction, Euclidean distance for color matching, and conditional GANs (cGANs) for outfit compatibility validation. Users provide natural language inputs specifying occasions and colors, which are converted into structured data for personalized recommendations. The architecture, combining data preprocessing, machine learning, and interactive modules, achieves a compatibility accuracy of 94.61%, precision of 92.82%, and perfect recall of 100.00% (F1-score: 96.27%). These results demonstrate the system’s reliability and potential to enhance user satisfaction in real-world applications.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Occasion and Color-Aware Personalized Outfit Recommendation System with Natural Language Interaction

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • B. K. Narasimha Shastry,
  • Ashish Lodaya,
  • Aditi Ponkshe,
  • Anjana Bharamnaikar,
  • Lalitha Madanbhavi,
  • Padmashree Desai,
  • Vijay Biradar

摘要

With the rise of e-commerce, demand for personalized fashion recommendation systems has increased. This study presents a novel approach that integrates occasion awareness, user-defined color preferences, and outfit compatibility testing using advanced machine learning and computer vision. The system employs K-means clustering for color extraction, Euclidean distance for color matching, and conditional GANs (cGANs) for outfit compatibility validation. Users provide natural language inputs specifying occasions and colors, which are converted into structured data for personalized recommendations. The architecture, combining data preprocessing, machine learning, and interactive modules, achieves a compatibility accuracy of 94.61%, precision of 92.82%, and perfect recall of 100.00% (F1-score: 96.27%). These results demonstrate the system’s reliability and potential to enhance user satisfaction in real-world applications.