Sturgeon AI

AI-assisted sex determination in sturgeon aquaculture

Status: Active Research (Grant Funded) Funding: ~$465,000 from USDA NIFA Western Regional Aquaculture Center Timeline: 2024-2027

Overview

This project develops artificial intelligence and machine learning approaches to revolutionize sex determination in sturgeon aquaculture. By combining computer vision and genomics, we aim to make sturgeon farming more economically viable and environmentally sustainable.

The Problem

Sturgeon aquaculture faces a critical challenge: females (caviar producers) cannot be reliably distinguished from males until 7-10 years of age. This leads to:

  • Massive resource waste feeding and housing males that will never produce caviar
  • Economic inefficiency with 50% of stock being unproductive
  • Environmental impact from unnecessary resource consumption

Our Approach

AI-Based Computer Vision

  • Non-invasive imaging of external morphology
  • Deep learning models trained on thousands of fish images
  • Real-time classification enabling rapid decision-making

Genomic Markers

  • Identification of sex-specific DNA sequences
  • Development of PCR-based diagnostic tests
  • Integration of genetic and phenotypic data

Current Results

  • Model accuracy improved from 76% to 90% for sex determination
  • Plans to expand dataset from hundreds to tens of thousands of images
  • Target: Sex detection earlier than three years of age

Expected Impact

  • Reduce operational costs by 30-50% through early culling of males
  • Lower carbon footprint of caviar production
  • Develop open-source computer vision framework for aquaculture

Collaborators

  • UC Davis - Department of Animal Sciences
  • University of Washington - Friday Harbor Laboratories
  • Industry Partners - Sturgeon growers in California and Idaho

Media Coverage