What is MediaMill?

The MediaMill team at the Intelligent Systems Lab of the University of Amsterdam aims to translate pixels to text for the purpose of retrieval.

A solution to visual search is highly relevant in a world that is adapting swiftly to visual communication via Internet and mobile phones. In contrast to search engines like Google, Facebook, and YouTube, which rely on what other people say is in the image, state-of-the-art visual search engines rely on an analysis of the image content. Such content-driven image search is crucial, if only to verify what people have said is factually in the images, pinpoint the exact location of query results within a video stream, or for professional archives which cannot be shared for crowdsourcing.

The research employed within MediaMill originates from various disciplines such as multimedia retrieval, image and video processing, computer vision, information retrieval, and machine learning. To ensure state-of-the-art competitiveness, MediaMill participates in the yearly TRECVID benchmark with its semantic video search engine.

Our visual search software is commercially available by sub-licensing via Euvision Technologies B.V..


MediaMill has its roots in the ICES-KIS Multimedia Information Analysis project (in conjunction with TNO) and the Innovative Research Program for Image processing (IOP). It blossomed in the BSIK program MultimediaN the EU FP-6 program VIDI-Video, the Dutch/Flemish IM-Pact BeeldCanon project, and the Dutch VENI SEARCHER project. The MediaMill team is currently funded by the Dutch VIDI STORY project, the Dutch FES COMMIT program, and the US IARPA SESAME project.

Latest News