Design of a Linked Outfit Recommendation System Based on Personal Gallery Data

Key Points

• This research aims to propose a system structure and user experience design for a personalized outfit recommendation service based on smartphone gallery data.
• Designed a system architecture utilizing gallery image recognition via ViT model.
• Implemented clothing attributes tagging using OpenCV.
• Integrated weather and schedule data for personalized recommendations.
• Defined a feedback-based learning mechanism for user experiences.
• Positive feedback on automatic registration of gallery images.
• Users found value in recommendations integrating temperature, perceived temperature, precipitation, and TPO.
• High usability and overall satisfaction were reported from the pilot evaluation conducted with 100 participants.