On January 8th, ReTV’s partners CERTH presented the ReTV-supported paper on “Unsupervised Video Summarization via Attention-Driven Adversarial Learning” at the 26th International Conference on Multimedia Modeling in Daejeon, Korea.

Authored by E. Apostolidis, E. Adamantidou, A. Metsai, V. Mezaris & I. Patras, the paper has been awarded the “Best Paper Award ” by the organizing committee of the conference.

The paper presents a new video summarization approach that integrates an attention mechanism to identify the significant parts of the video, and is trained unsupervisingly via generative adversarial learning. 

Are you interested in knowing more? You can read the full-text paper here, access the presentation slides here and the related software used in the paper: here.