Abstract
After almost three years of development, the Video Hunter tool (formerly the Signature-Based Video Browser) has become a complex tool combining different query modalities, multi-sketches, visualizations and browsing techniques. In this paper, we present additional improvements of the tool focusing on keyword search. More specifically, we present a method relying on an external image search engine and a method relying on ImageNet labels. We also present a keyframe caching method employed by our tool.
Cite
CITATION STYLE
Blažek, A., Lokoč, J., & Kuboň, D. (2017). Video hunter at VBS 2017. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 10133 LNCS, pp. 493–498). Springer Verlag. https://doi.org/10.1007/978-3-319-51814-5_47
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
