Automated Image-based Taxon Identification Using Deep Learning And Citizen-science Contributions

This thesis explores the potential of leveraging recent technological innovations to accelerate taxonomic research, which is a critical bottleneck in the current sixth mass extinction. The proposed solution utilizes a technique called feature transfer, in which a pretrained convolutional neural network (CNN) is used to obtain image representations (“deep features”) for a taxonomic task of interest. These features are then used to train a simpler system, such as a linear support vector machine classifier. The results of the research show that this technique is reliable, inexpensive, and generally applicable, enabling taxonomists to build their own automated identification systems without prohibitive investments in imaging and computation.