Bringing Computer VISION into the Classroom Utilizing RCA Imagery and the OOI Jupyter Hub

There is a rapidly growing demand in Earth system science for workforce expertise in machine learning. To increase marine science students understanding of, and ability to use, artificial  intelligence tools, Dr. Katie Bigham and School of Oceanography undergraduate student Atticus Carter are teaching an undergraduate class focused on applying “computer vision” (processing imagery with the computer) to marine science problems. The Computer Vision Across Marine Sciences prototype course for UW undergraduates in the Ocean Technology program includes the development of a Jupyter Binder (a notebook collecting multiple Jupyter Notebooks). Currently, an enormous amount of time is required to manually process imagery collected in marine environments, resulting in a major bottleneck for all research utilizing marine imagery. In this course, students gain an understanding of computer vision capabilities, model training and evaluation, and research applications. The course utilizes real-world datasets from the OOI Regional Cabled Array (RCA), including imagery from remotely operated vehicles utilized on RCA cruises and fixed camera imagery on the array, and from other systems, exposing students to diverse marine habitats. For their final projects, students will develop bespoke models utilizing datasets of their choosing, including RCA imagery. Students can employ the OOI Jupyter Hub for additional computational power, facilitating easy access to imagery and necessary resources. In collaboration with a faculty member in the UW School of Education quantitative data are collected on student learning and feedback for course improvement. The open-access course text is actively under development and is accessible at OceanCV.org. The team aims to improve the materials based on student feedback. Carter will present findings and learnings from the first version of the class at ASLO 2025 Aquatic Sciences Meeting in the Building Data Literacy Skills in the Next Generation of Aquatic Scientists session hosted by OOI Data Labs.

Figure 27: Introductory page for Computer Vision Across Marine Sciences Jupyter Binder and example of artificial intelligence-based predictions of animals in imagery collected by the Regional Cabled Array digital still camera at Southern Hydrate Ridge. This work is supported by an NSF OCE Postdoctoral Research Fellowship to Dr. K. Bigham, University of Washington.