Student intern: Tate Dunbar (Computer Science, Biology)
Supervisors: Fergus Chadwick (School of Mathematics and Statistics), Kasim Terzic (School of Computer Science)
Citizen science involves non-professionals participating in research. Data from the public is increasingly used in ecological and conservation studies, to increase the range of data available to researchers, particularly in the field of ornithology (e.g. the Cornell Lab survey sites). However, the risk of such data collection methods is that less expert observations may lead to inaccurate data collection.
Convolutional neural networks (CNNs) are a form of AI that can be trained to recognise visual images. The aim of this study was to investigate the performance of several CNNs on a bird species classification tasks, to identify the best performing one, and to use it to try to understand how accurate classification is achieved. Understanding the decision-making processes involved may help us to better understand how humans might classify birds with a lower rate of error.
