In a groundbreaking development for the aquaculture industry, researchers from UC Davis and the University of Washington are leveraging advanced AI and high-resolution imaging to revolutionize sex determination in juvenile sturgeon. This innovative project, led by UC Cooperative Extension Specialist Jackson Gross and computer scientist Edwin Solares, aims to create a cost-effective, scalable, and non-invasive solution for sturgeon producers.
The journey began with a proof-of-concept study, where Solares and Gross collaborated with California sturgeon producers to collect preliminary images. The goal was to train AI models to identify subtle anatomical differences between male and female sturgeon, which are virtually indistinguishable to the human eye, even for experts. Past research has suggested that juvenile sturgeon exhibit subtle but detectable sexual dimorphism in their ventral anatomy, and the team is building on this foundation.
The initial AI models achieved an impressive 76% accuracy, but the researchers are not stopping there. With continued support from the USDA NIFA Western Regional Aquaculture Center, they are expanding their dataset to tens of thousands of images. This will allow them to refine their models and improve accuracy, with the ultimate goal of detecting sex in sturgeon younger than three years old.
The team is utilizing the Expanse system at the San Diego Supercomputer Center (SDSC) to leverage machine learning advancements. By accessing high-performance computing resources through the NSF ACCESS program, they can process and analyze vast amounts of data efficiently. This will enable them to develop a user-friendly, farm-level solution that can be seamlessly integrated into daily operations.
Adam Summers, a professor at the UW Friday Harbor Laboratories, emphasized the practicality of their approach. “Our software will run on a mobile platform, receive over-the-air updates, and require minimal training for farm staff,” he said. This ensures that the technology is accessible and easy to use, even for those without extensive technical expertise.
In addition to developing a practical solution, the research team will conduct a comprehensive review of existing AI applications in aquaculture. By showcasing their innovative approach through high-resolution imagery and detailed algorithmic analysis, they aim to provide a transparent and scientifically rigorous demonstration of their methodology.
The implications of this research extend beyond the sturgeon industry. As Summers noted, “Our findings will not only benefit sturgeon producers but also serve as a model for AI applications in aquaculture more broadly.” This innovation could set a new standard for non-invasive, efficient fish management, paving the way for further technological advancements in the field. Moreover, it could aid in conservation efforts for endangered species of sturgeon and potentially other fish species.
The upcoming research is funded by the USDA NIFA Western Regional Aquaculture Center, with computational resources provided by the U.S. NSF ACCESS program. As the project progresses, the team remains committed to transparency and scientific rigor, ensuring that their findings are both reliable and impactful. The aquaculture industry is on the cusp of a technological revolution, and this research is leading the charge.
