In this chapter we highlight how rapid advances in computer vision and the increasing availability of high-resolution satellite imagery have facilitated more accurate, efficient, and scalable environmental monitoring and regulation. First, we highlight the range of potential use cases of remote sensing with satellite imagery in environmental enforcement. Second, we describe the methodological evolution from manual learning from satellite imagery, to model-based inference largely based on pixel-by-pixel classification, to deep learning. Third, we provide an in-depth case study, illustrating how deep learning with satellite imagery can solve a problem that has vexed the Environmental Protection Agency for decades: the identification of Concentrated Animal Feeding Operations (CAFO), which pose substantial environmental risk. Last, we highlight the data infrastructure, modeling, and capacity challenges that must be overcome to facilitate this profound shift in the evidence base for environmental enforcement.