Fetching the paper…

Rekall: Specifying Video Events using Compositions of Spatiotemporal Labels · Around