In this project, I explore how can we can get the most out of pre-trained computer vision models and how to optimize them efficiently under compute constraints. It serves as a foundation to my black box image augmentation project.

This work has been published on the MLConf blog.