Significant strides have been made within computer vision (CV) especially through developments leveraging deep learning. The applications are numerous but each use case comes with unique challenges as well as those that are common. For instance, safety critical applications of passive imaging combined with current CV approaches are too brittle, lack sufficient explainability, are difficult to manage, and are extraordinarily greedy. This talk will explore the issues, the demands, and potential routes forward to allow us to expand the utility of computer vision to more applications and more sectors.