Minor Area Motion Imagery (MAMI) Dismount Tower Data challenge problems

Abstract

The ability to classify a dismount and its activity, is of interest for both military and non-military applications. This effort describes a database that is rich for dismount activity classification and is available to the public - the Minor Area Motion Imagery Dismount Tower Data (MAMI-DTD) Collection. The MAMI-DTD collection was gathered in the Summer of 2013 and contains several examples of dismount activity such as running, walking, walking with a load, etc. It is unique because it contains a variety of operating conditions including angular diversity. Furthermore, it contains multi-modal data - infared (IR), passive visible, etc. This paper provides a detailed description of the data collection and details some interesting challenges such as gender, activity, and hierarchical classification.