As in previous years, also the 9th Video Browser Showdown (VBS 2020) was a great success and a wonderful event. Even though we used a very challenging dataset (V3C1) with 1000 hours of video content, all teams were able to solve a lot of search tasks very quickly. While being a professional video retrieval evaluation competition, it was a very exciting and entertaining event as well (see pictures below)! We thank all participants and contributors and look forward to the next iteration of the VBS, which will take place in January 2021 in Prague, Czech Republic (CFP will follow). New teams that want to join should contact us as soon as possible, in order to share data, hints, and maybe even some tools that can help to start! Further down below you can also find examples of VBS systems, which will give you some inspiration for your own system.
The VBS 2020 took place in Daejeon, South Korea, on January 6, 2020. We used 1000 hours of video content for the evaluation (V3C1 dataset) and a lot of challenging visual and textual KIS tasks.
The VBS lasts for a few hours, where either experts (developers/researchers) or novices (volunteers from the audience) of all teams try to solve KIS and AVS tasks as fast as possible.
The VBS 2019 took place from January 8-9 2019 at MMM 2019 in Thessaloniki, Greece. We evaluated KIS and AVS tasks on a large, shared video dataset (V3C1 - Part of the Vimeo Collection with Creative Commons license).
Known-Item Search (KIS) require to find a very specific segment in the video collection (which is either described by text or video presentation), while Ad-Hoc Video (AVS) task require to find as many examples of a specific topic, as possible.
The video retrieval tool from the SIRET research group at Charles University in Prague (Czech Republic) was the overall winner at the VBS 2018 competition.
This is very flexible interactive video retrieval tool developed by the Institute of Information Technology (ITEC) at Klagenfurt University (Austria). At VBS 2018 it scored 2nd in total.
This is the video browsing system developed by HTW Berlin (Germany) and used for the VBS for several years with incremental improvements. In 2016 this tool won the VBS competition!
This tool is follow-up development of the IMOTION search system, which won the VBS 2017 competition. It provides many different and effective filtering options with efficient search methods and is available as an open source software!