NOAA Puget Sound Nearshore Fish 2017-2018

Published: 4 August 2023| Version 1 | DOI: 10.17632/n73g6ysv8c.1
Dara Farrell,
, Anusua Trevidi, Shreyaan Pathak, Sreya Muppalla, Jane Wang, Dan Morris, Rahul Dodhia


The sustainable management of fisheries and aquaculture requires an understanding of how these activities interact with natural fish populations. GoPro cameras were used to collect an underwater video data set on and around shellfish aquaculture farms in an estuary in the NE Pacific from June to August 2017 and June to August 2018 to better understand habitat use by the local fish and crab communities. Images extracted from these videos were labeled to produce a data set that is suitable for use in training computer vision models. This approximately 8 GB data set contains 77,739 images sampled from the collected videos; 67,990 objects (primarily fishes and crabs) have been annotated in 30,384 images. The remainder of the images (47,355 images) have been annotated as “empty”. All images (with and without animals) are at a resolution of 1920 x 1080. Annotations have been provided in the COCO Camera Traps JSON format. Additional information can be found here: For particular ecological studies, such as studies of animal behavior, it is desirable to minimize changes to the natural environment when collecting data, which discourages the use of artificial light sources. Data sets that feature brackish water are rare, and naturally-lit examples showcasing field data collection challenges are even more rare, and will allow for the training of models that are able to support such studies. These data have the potential to help researchers address system-level and in-depth regional shellfish aquaculture questions related to ecosystem services and shellfish aquaculture interactions.



University of Washington, National Oceanic and Atmospheric Administration


Artificial Intelligence, Aquaculture, Object Detection, Machine Learning