Efficient Thumbnail Identification through Object Recognition
Year: 2020
Type: conference-paper
Abstract: Given the overwhelming growth of online videos, providing suitable video thumbnails is important not only \nto influence user’s browsing and searching experience, but also for companies involved in exploiting video \nsharing portals (YouTube, in our work) for their business activities (e.g., advertising). A main requirement for \nautomated thumbnail generation frameworks is to be highly reliable and time-efficient, and, at the same time, \neconomic in terms of computational efforts. As conventional methods often fail to produce satisfying results, \nvideo thumbnail generation is a challenging research topic. In this paper, we propose two novel approaches \nable to provide relevant thumbnails with the minimum effort in terms of time execution and computational \nresources. The proposals rely on an object recognition framework which captures the most topic-related \nframes of a video, and selects the thumbnails from its resulting frames set. Our approach is a trade-off between \ncontent-coverage and time-efficiency. We perform preliminary experiments aimed at assessing and validating \nour models, and we compare them with a baseline compliant to the state-of-the-art. The assessments confirm \nour expectations, and encourage the future improvement of the proposed algorithms, as our proposals are \nsignificantly faster and more accurate than the baseline.
Institution: University of Cagliari
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W3103589845