SBIR/STTR Award attributes
The 412th Test Wing CEG/CEVA has a national defense-related mission need in the area of Computer Vision Machine Learning for Biological Species Identification. Specifically, biological surveys are often conducted using fixed-location, motion-sensing digital cameras in order to determine the status of species. These camera surveys may generate millions of photographs that must all be individually analyzed leading to large cognitive load. This analysis is time consuming because a person must view every photograph, and inefficient because most photographs do not contain the species of interest. In addition, photographs may include important species other than the species of interest, and this information may be lost due to time constraints. This project would utilize computer vision and machine learning to identify all plant and animal species in each photograph and to catalog the photographs appropriately. The identification would ingest AF specific ontologies depending on where the system is deployed in the world and also be hierarchical to maximize the likelihood of accurate and useful information. For instance, the Mohave Ground Squirrels (MGS) could be classified mammal/rodent/ground squirrel/Mohave ground squirrel/sex/age class/individual. Similar hierarchies would be developed for other plant and animal species. Successful implementation would significantly decrease the cost of conducting biological surveys.
