Comparative Performance Analysis of Image Classification Algorithms on SEMs of the Bryozoan Order Cheilostomata

With Arthur Porto, Florida Museum of Natural History, Department of Biovision 

September 2024 - 

 

Image classification algorithms play a crucial role in various fields, and they could be a great asset for taxonomists to accelerate identification of specimens. However, the performance of these algorithms trained on biological datasets, remains largely unexplored. Conventionally, these models are trained on large databases of millions of images. Biological datasets, on the other hand, often suffer from imperfections, such as insufficient data for understudied or rare taxa.

​

This study aims to compare the performance of five different image classification algorithms on scanning electron microscope images of Cheilostomata and determine which algorithm achieves the highest accuracy in identifying SEMs to genera within this taxonomic group.

 

My responsibilities include cleaning up and expanding the database, training AI models on the image dataset, assessing their performance, and collaborating with researchers to refine the algorithms. This project allows me to apply my computational skills to solve complex taxonomic problems, bridging the gap between technology and biology.