Automated identification of bees

Biodiversity is an important indicator of ecosystem health and functioning. Assessing biodiversity necessitates surveying insects within that ecosystem, but identifying the species is a daunting task because of the skill sets required and the time needed to develop them. Automated identification systems, such as SPIDA, can reduce the time necessary for identification because it does not rely on user knowledge. Preliminary work has shown that this system can be used to identify at least six species of North American bees.  This project develops a fully automated program that uses image processing and an Artificial Neural Network (ANN) to quantitatively identify bees to species level.  For each species of bees used, we entered sets of forewing images into the program to train and test the system.  Because this program only requires the input of bee forewing images by the user, it makes taxonomy more accessible and greatly reduces the time required for identification.  With this automated system, we can in effect use pre-existing collections of identified bees to create a global key, which could allow for quicker detection of invasive species and assessment of overall biodiversity health.