Timezone: »

 
Self-Contained Entity Discovery from Captioned Videos
melika ayoughi · Paul Groth · Pascal Mettes

This paper introduces the task of visual named entity discovery in videos without the need for task-specific supervision or task-specific external knowledge sources. Assigning specific names to entities (e.g. faces, scenes, or objects) in video frames is a long-standing challenge. Commonly, this problem is addressed as a supervised learning objective by manually annotating faces with entity labels. To bypass the annotation burden of this setup, several works have investigated the problem by utilizing external knowledge sources such as movie databases. While effective, such approaches do not work when task-specific knowledge sources are not provided and can only be applied to movies and TV series. In this work, we take the problem a step further and propose to discover entities in videos from videos and corresponding captions or subtitles. We introduce a three-stage method where we (i) create bipartite entity-name graphs from frame-caption pairs, (ii) find visual entity agreements, and (iii) refine the entity assignment through entity-level prototype construction. To tackle this new problem, we outline two new benchmarks SC-Friends and SC-BBT based on the Friends and Big Bang Theory TV series. Experiments on the benchmarks demonstrate the ability of our approach to discover which named entity belongs to which face or scene, with an accuracy close to a supervised oracle, just from the multimodal information present in videos.Additionally, our qualitative examples show the potential challenges of self-contained discovery of any visual entity for future work.

Author Information

melika ayoughi (university of amsterdam)
Paul Groth (University of Amsterdam)
Paul Groth

Paul Groth is Professor of Algorithmic Data Science at the University of Amsterdam where he leads the Intelligent Data Engineering Lab (INDElab). He holds a Ph.D. in Computer Science from the University of Southampton (2007) and has done research at the University of Southern California, the Vrije Universiteit Amsterdam and Elsevier Labs. His research focuses on intelligent systems for dealing with large amounts of diverse contextualized knowledge with a particular focus on web and science applications. This includes research in data provenance, data integration and knowledge sharing.

Pascal Mettes (University of Amsterdam)

More from the Same Authors