Current video retrieval efforts all found their evaluation on an instance-based assumption, that only a single caption is relevant to a query video and vice versa. In this paper, we demonstrate this assumption results in performance comparisons often not indicative of models’ retrieval capabilities. We propose a move to semantic similarity video retrieval, where (i) multiple videos can be deemed equally relevant, and their relative ranking does not affect a method’s reported performance and (ii) retrieved videos (or captions) are ranked by their similarity to a query. We propose several proxies to estimate semantic similarities in large-scale retrieval datasets, without additional annotations. Our analysis is performed on three commonly used video retrieval benchmark datasets (MSR-VTT, YouCook2 and EPIC-KITCHENS).
Find out more: [ Ссылка ]
Read the paper: [ Ссылка ]
Visit the GitHub Repo for code: [ Ссылка ]
