A Study of Performance Improvement for Video Retrieval System Based on Relevance Feedback

Recently, an effective retrieval system for video data based on its content is required. In such system, queries are given by natural video data or drawing sketches. However, preparing natural video data and drawing sketches require high operating costs for users. Therefore, we proposed novel method that uses "Relevance Feedback(RF)" for creating query. RF is an approach that is used in general information retrieval. RF allows us to give a query to the system by simple operation. Then, we performed an experiment to retrieve video data by using the system that uses the method. The experimental results lead us to conclude that the system is effective at reducing a user’s operating costs. Furthermore, we show the future direction for research.