Our goal is to create a comprehensive inventory management system using Deep Learning-backed Computer Vision methods. With Computer Vision, we will implement object recognition of essential inventory items on board the Howitzer, allowing for updated, accurate counts in real-time. Using Deep Learning will create an adaptable system that is robust to high levels of noise and occlusion we anticipate from imaging in a combat setting. In Phase I, we aim to demonstrate the feasibility of creating a system capable of object recognition for an item with similar characteristics to a critical item, expand to all items. We will create a system that displays active inventory, provides low-stock warnings, and interfaces with other automation efforts.