Efficiency In Computer Vision: From Compute And Memory To Robustness