A High-Throughput Automated Microfossil Classification System Using Deep Learning

The rapid advancement of computational power has facilitated the widespread adoption of deep learning, a subset of artificial intelligence (AI), in various fields. Automated microfossil classification using AI is increasingly explored as a solution to reduce labor and address the declining availability of skilled personnel. However, practical implementation in research remains limited due to challenges such as the need for extensive training datasets and the lack of advanced equipment like automated microscopes. To address these issues, we implemented deep learning as a function to automatically classify microfossils on a virtual slide scanner that can process up to 360 microscope slides continuously. This study applied the system to sediment core DCR-1PC from the Indian Ocean sector of the Southern Ocean to obtain high-resolution records of the radiolarian analysis.