Fisheries and Oceans Canada / Pêches et Océans Canada

This project developed a machine learning-based method to identify environmental extremes by segmenting data into regions with consistent conditions (e.g., temperature, oxygen, acidity).

This project applied neural networks to post-process streamflow outputs from the WRF-Hydro model, improving alignment with observational data.

GitHub Copilot is being piloted at DFO aimed at enhancing developer productivity and exploring the operational value of AI-assisted coding.

We are looking into using Labels back end for smart labels as well as a free copilot agent for the department to use in order to validate a user's classification.

This completed project used Random Forest and Convolutional Neural Network (CNN) algorithms trained on manually identified fish sounds for detection in passive acoustic recordings.

Problem: DFO-Sclerochronlogy Lab is responsible for fish age analysis services to clients across British Columbia and the Yukon and is responsible for approximately 80,000 salmon scale aging assessments a year.

Problem: Information detailing DFO policy and governance, science and research, and main activities in BC and the Yukon for management of fisheries are offered on the DFO Pacific Fisheries internet site.

Problem: DFO inspects hours of underwater camera footage to count and identify fish species through fence counts, which is a structure installed across a river or stream to monitor fish population.

Problem: Efficiently extracting and navigating information from machine-readable PDFs.

Problem: Digitizing handwritten or non-machine-readable documents using optical character recognition (OCR) often introduces transcription errors, which can compromise data quality and usability.

Problem: Identifying emerging topics that are relevant to DFO's mandate.

Problem: Science observations and research requires transcription and validation from hand-written field notes, including historical information.