Pascal Mettes and Cees Snoek from the Informatics Institute were awarded at last month’s ACM Multimedia conference. Pascal Mettes was voted best doctoral student and together with researchers from Zhejiang University and Renmin University Cees Snoek won the first prize in the Multimedia Grand Challenge.
To automatically recognize objects, scenes and actions in videos with artificial intelligence, large amounts of labeled examples are a necessity. Pascal Mettes explores in his thesis how limited amounts of human supervision may still achieve precise automatic results. In an international field of candidates Pascal presented the progress of his thesis, showcased his recent articles and convinced experts of his approach.
Summarization of a video clip into its most descriptive sentence was the Multimedia Grand Challenge at the ACM Multimedia conference. Together with researchers from China, Cees Snoek managed to best meet this challenge. Using deep learning they managed to develop a system that automatically translates video clips to text. On a Microsoft Research-defined video collection the sentences that the system generated resembled human-provided summaries the most.
The ACM International Conference on Multimedia is the premier conference in the multimedia field, which publishes high quality papers presenting novel theoretical and algorithmic solutions addressing problems across the domain of multimedia and related applications. The 2016 edition was held in movie theater Tuschinski in Amsterdam.