Crowd Counting and Localization for Subsurface Operations

Accurate crowd counting and localization are particularly important in subsurface environments to assist decisionmakers in critical security situations. Thus, we introduce an approach based on artificial intelligence to count and localize individuals, both emergency personnel and uninvolved persons, to resolve crowded situations. Our method needs a single image to predict the overall crowd count, density map, and precise locations of each individual, which could then be mapped to a common operational picture. With the additional precise person location information more concise end user information can be made possible, while simultaneously improving absolute counting information.