no code implementations • ECCV 2018 • Pouya Samangouei, Ardavan Saeedi, Liam Nakagawa, Nathan Silberman
We introduce a new method for interpreting computer vision models: visually perceptible, decision-boundary crossing transformations.
