1 paper · 1 filter
Kevin D. Nguetche, Lilian Hollard, Eric Granger +1
Many temporal approaches for video recognition rely on explicit cross-channel aggregation projections, which have quadratic complexity in the number of channels, introducing signif…