Data-Driven Curated Video Catalogs: Republishing Video Footage

This project illustrates data-driven curated video catalogs as an approach for the analysis of video footage. Having rich and diverse collections of videos as inputs, data-driven catalogs seek to identify common objects and reorganize them into thematic clusters displayed in a video format. The technique takes inspiration from two scientific practices (core sampling and light diffraction) and two publishing formats (supercuts and visual catalogs). Data-driven curated video catalogs are used to republish a collection of found footage of 2019 Veniceʼs high water that devastated the city in an unprecedented way. Starting from an editorial selection of footage culled from YouTube, various algorithmic processes are used to demarcate and reorganize the material into thematic video series (people, boats, and birds). The resulting video catalogs support a type of visual analysis that goes beyond traditional forms of measurement while, at the same time, carrying an expressive power.