Now live on Zooniverse! Explore Washington’s underwater forests and contribute directly to conservation research!
Check out our project update to learn more about the images you are seeing.
Important note: Subjects from Workflow 1 (Yes/No) are routed into Workflow 2 (multiple choice). This means Workflow 2 may not always have images available. Please check back regularly to continue contributing to the multiple choice workflow. Thank you!
Results
Open-source, open-access research at the Seattle Aquarium
To enable transparency and reproducibility, we are committed to open-source and open-access research. “Open source” means the underlying computer code we use—such as software and machine learning (ML) model frameworks—is publicly available for anyone to inspect, use and improve. “Open access” means we make our research outputs, including trained models and documentation, publicly available whenever possible.
We also rely on the products of entities who share our value of accessibility in science. We utilize the robust YOLO11-cls model framework developed and freely provided by Ultralytics to train our image classifier model. Specifically, we use CoralNet-Toolbox, created by Jordan Pierce, the creator of the Zooniverse project Click-a-Coral, to implement this model. Because the framework is open source, we can adapt and retrain it for kelp forest imagery, and other researchers can apply the same approach to their own datasets or build upon our methods. For more information about the open-source ML model we use (YOLO11s-cls) and how it works, please see the FAQ section.
Our commitment to openness extends beyond the ML model. We also share our ROV survey methods, data-processing workflows and analytical approaches so that other groups can replicate our methods, learn from our lessons and adapt them to their own coastal monitoring programs.
With your help improving its predictions, our trained model will become a powerful tool for processing large volumes of kelp forest imagery. By keeping both our methods and models accessible, we support independent validation, reuse in new regions and broader applications in kelp forest conservation and monitoring.
Below, we list multiple websites and resources where people can learn more, including viewing areas of active field and analytical development (often via the “Issues” tab on GitHub repositories).
ROV imagery and media
- Linked here is a folder of ROV survey imagery featuring the diverse algae and invertebrates of Elliott Bay, Washington, U.S. These are the types of images you will help us review as part of Kelp Quest!
- Linked here is a full transect (a discrete ROV survey) containing 75 photos. This was a 30-meter ROV transect offshore of Centennial Park, Seattle, Washington, U.S., flown at a depth of 8 meters (26 feet) beneath the sea surface, with the ROV 0.8 meters (2.62 feet) above the seafloor.
- If you want to dive deep into our ROV field methodology, statistical analyses and our machine learning methods of data extraction that generate the subjects you will review, please download the report linked here.
- To access a large set of "raw" and "post-processed" images used to train machine learning models to process our survey imagery, we have made one available on GitHub, linked here.
- To watch a video of a “day in the life” of the ROV research team, click here.
- To see a video featuring our collaboration with the Port of Seattle, click here.