Each extreme natural disaster is an opportunity to assess the performance of our infrastructure under circumstances that cannot entirely be reproduced in the laboratory or through numerical simulation. When performing reconnaissance in the field, engineers record much of the information being gathered in the form of images. When large quantities of images are collected within a short time period and have such a wide variety of content, timely organization and documentation will be important. Clearly, manual sorting and analysis of all these images would be prohibitively tedious and expensive. Thus, to efficiently and thoroughly use these images, engineers need to exploit recent advances in computer vision and machine learning. In this project, we devised a series of automated capabilities, based on computer vision and machine learning, to classify, organize, and document the large volumes of visual data collected during a reconnaissance mission. We have developed a suite of techniques that can significantly enhance current data collection and organization procedures to support reconnaissance missions.