Deep-Sea Biodiversity: VIAME and OOI JupyterHub Accelerate Research

By Amber Stronk | September 20, 2024 |

Off the coast of Oregon, the Axial Seamount on the Juan de Fuca Ridge hosts the ASHES vent field, a site rich in hydrothermal activity and diverse marine life. Principal Investigators, Dax Soule and Karen Bemis, and their teams are leveraging footage from past expeditions, captured by the OOI CamHD video archive, to study species diversity and abundance in this extreme environment.

With hundreds of terabytes of video data available, manual analysis is impractical. To address this, master’s student Julia Sandke and a team of undergraduate researchers are utilizing VIAME (Video and Image Analytics for a Marine Environment), a tool that helps build machine learning datasets by automating the annotation and classification of vent organisms. The goal is to use these datasets to train models that can autonomously analyze species diversity and abundance.

The integration of VIAME’s directories into the OOI JupyterHub environment, along with the addition of Nvidia L40S GPUs, has been a pivotal advancement. Previously, the team faced long delays using the shared VIAME web interface. Now, with dedicated resources in JupyterHub and the enhanced power of GPUs, they can significantly speed up machine learning processes, such as video training and data analysis, which once took days to complete.

This new computational power allows the team to begin unlocking the wealth of data in the CamHD archive, delving deeper into the species diversity around these vents. The long-term goal of the project is to use computer vision and machine learning to develop models that can track species abundance and predict key changes in the ecosystem. For example, these models could forecast where target species may appear after volcanic eruptions at Axial Seamount, providing researchers crucial insights into the shifting biological landscape of hydrothermal vent environments.