1 An Effective Video Retrieval System by Combining Visual and Textual Mining Techniques
Ja Hwung Su, Hsin-Ho Yeh, Shih-Pang Tseng · 2016
The prime goal of video retrieval systems is to facilitate the users in finding the desired video contents. Nowadays, the most popular way is textual-based search engine, such as Youtube, Yahoo, etc. However, high-priced manual annotation cost makes it inefficient in preprocessing the metadata. Another solution for video retrieval is visual-based retrieval system. Apart from the metadata, the retrieval quality is still tied to the visual-diversity problem. To overcome the problems existing in the individual approach, in this paper, we propose a new system by combining textual- and visual-based mining techniques. Based on the proposed visual- and textual-based retrieval models, the video retrieval can be achieved effectively. Through empirical evaluations, the proposed system can successfully associate the target videos and the user’s interest. Index Terms — content-based video retrieval, textual-based video retrieval, multimedia mining, temporal pattern mining