We used Adaptive Resolution Imaging Sonar (ARIS), which is based on Dual-Frequency Identification Sonar (DIDSON) technology to observe net encounter behaviors of bigheaded carps. Gill nets with four different mesh types (8.9 cm bar mesh) and trammel nets with five different mesh sizes (range 5.1-8.9 cm bar mesh) were evaluated. Net sets were undisturbed, or fish were driven with the noise of the outboard boat motor. We sampled field sites in tributaries and the mainstem Missouri River in central Missouri from June 2015 to October 2016. Specific sampling sites were determined based on previously documented Silver Carp abundance with anticipated densities sufficient for testing behavioral responses to entanglement gears. Specific tributaries sampled consisted of the Lamine River, Blackwater River, Moniteau Creek, and Cedar Creek. Habitat within these low gradient tributaries consists of minimal current if present and shallow depths (less than 4 m). Nets set in the mainstem Missouri River were downstream of channel training structures, where flow diversion resulted in minimal to no current and similar depths. Eighty net sets for 40 hours of total ARIS video were collected. Catch was used to identify length and species present in videos, with Silver Carp and Smallmouth Buffalo the predominate species caught. Within ARIS videos, body morphology was used to categorize and enumerate responses of fish.

This dataset includes post-processed data for underwater fish observations by using Acoustic Resolution Imaging Sonar technology in Montezuma Slough, Sacramento River, and in the Sacramento Deep Water Shipping Channel in the Sacramento-San Joaquin Delta, California, USA. This data release includes sample data associated with each video file collected, swimming behavior of groups of released cultured Delta Smelt (Hypomesus transpacificus) that is associated with all three release types, and tracking data on both released cultured Delta Smelt and ambient fish that were observed underwater using sonar technology. This data release includes data from years one to three of the Delta Smelt Experimental Releases, in water years 2022 to 2024.

In September 2023, the USGS demonstrated the use of side-scan sonar to image large densities of invasive carp in Creve Coeur Lake, Maryland Heights, Missouri, USA. A 400-meter reach was surveyed using a Humminbird Helix sonar system. Proprietary sonar data were converted to PNG image files (i.e., rectified and speed-corrected) using SonarTRX Pro. The resulting data set includes one CSV file, original side-scan sonar images (PNG files), speed corrected side-scan images (correcting distortions in imagery due to small changes in boat velocity; PNG files), and a KMZ file of original rectified images for a mosaic display in Google Earth Pro.

The global seafood production of marine finfish is dominated by Atlantic salmon (Salmo salar) reared in marine net-pens. Incidents and structural failures may lead to salmon escaping net-pens with their fate being poorly understood, particularly for small-scale escape events (N < 100). This study sought to identify spatial patterns and the fate of Atlantic salmon post-smolts following a simulated escape event from an aquaculture site in the Bay of Fundy. Farmed salmon (N = 99) were implanted with an ultrasonic acoustic tag and were released from a farm in small groups (N ≤ 5). Using an established acoustic array, we tracked fish movement patterns and mortality over a ~4 month period. Predation events by warm-bodied predators were assayed using the tag’s temperature sensors. Fish took ~5.0 h to leave the release site and initially opted to either move further inshore (N = 8) or out to sea (N = 73), though a large number of fish returned to the release site spending a large proportion of their time budget there (~26.4%). Several fish (N = 14) were also observed in nearby river estuaries. Most fish (72.7%) succumbed to predation shortly after release (mean time to predation = 57.9 h). Our findings suggest that there is a strong association of salmon with the release site that may result from seasonal migratory tendencies, which could be exploited to improve recovery of escapees. Also, predation may serve to limit the number of escapee salmon entering the natural environment following a small-scale release. Cite this data as: Lawrence, M.J., Wilson, B.M., Wringe, B.F., Hawkes, J.P., Hardie, D.C., Hamoutene, D., Flávio, H., English, G., Black, M., McKindsey, C.W., Trudel, M. Movement patterns and predator-prey interactions of domestic Atlantic salmon (Salmo salar) following an experimental release in a highly dynamic marine environment. Published April 2025. Coastal Ecosystem Science Division, Fisheries and Oceans Canada, St Andrews, NB.

In January 2023, the USGS demonstrated the use of side-scan sonar and down-imaging sonar to locate and quantify fish including invasive carp. A dike pool in the study area was surveyed (254-meter reach) using a Humminbird Helix sonar system. Sonar surveys included concurrent recordings of side-scan and down-imaging sonar. Proprietary sonar data were converted to PNG image files (i.e., rectified and speed-corrected) using SonarTRX Pro. Fish targets in imagery were measured in total length and enumerated using image analysis software (ImageJ) algorithms. Side-scan and down-imaging datasets were processed multiple times across a range of sensitivity thresholds (i.e., 25–30) to incorporate variation in the sensitivity of fish detection and estimation. The resulting data set includes three CSV files documenting the track locations and beam information needed to calculate the ensonified volume and the fish counts and length of each fish captured in each sonar type.

The Fish Detection AI project aims to improve the efficiency of fish monitoring around marine energy facilities to comply with regulatory requirements. Despite advancements in computer vision, there is limited focus on sonar images, identifying small fish with unlabeled data, and methods for underwater fish monitoring for marine energy. A Faster R-CNN (Region-based Convolutional Neural Network) was developed using sonar images from Alaska Fish and Games to identify, track, and count fish in underwater environments. Supervised methods were used with Faster R-CNN to detect fish based on training using labeled data of fish. Customized filters were specifically applied to detect and count small fish when labeled datasets were unavailable. Unsupervised Domain Adaptation techniques were implemented to enable trained models to be applied to different unseen datasets, reducing the need for labeling datasets and training new models for various locations. Additionally, elastic shape analysis (ESA), hyper-image analysis, and various image preprocessing methods were explored to enhance fish detection. In this research we achieved: 1. Faster R-CNN for Sonar images - Applied Faster R-CNN reached > 0.85 average precision (AP) for large fish detection, providing robust results for higher-quality sonar images. - Integrated Norfair tracking to reduce double-counting of fish across video frames, enabling more accurate population estimates. 2. Small Fish Identification - Established customized filtering methods for small, often unlabeled fish in noisy acoustic images. This submission of data includes several sub-directories: - FryCounting: contains information on how to count small fish (i.e., fry) in the sonar image data - SG_aldi_addons: contains additions to the ALDI code (SG = Steven Gutstein, primary author) such as the trained models used in this experiment, which should match the models achieved when the training instructions are followed, and code for how to make the sonar images into movies - Summaries_Dir: contains information on how to set up the foundation to perform these experiments, such as installing all required packages and versions, and creating the PyTorch and ALDI environments These experiments boil down to a 2-part structure as described in the uploaded readme file: Part I: Installing and Using ALDI & Norfair Code - This is used for tracking and counting fish, and is a replication of the article that is linked, namely the Align and Distill (Aldi) work done by Justin Kay and others - This part relates to the Summaries_Dir subfolder, and the SG_aldi_addons sub-folder Part II: Installing and Using Fry Code - This is used to track and count smaller fish (aka fry) - This relates to the FryCounting sub-directory Also included here are links to the downloadable sonar data and the article that was replicated in this study.

This project is a collaboration between the Department of Fisheries, Murdoch University and Curtin University and the University of WA. It will use the latest sonar and video technologies to study spawning aggregations of fish off the west coast, particularly samson fish, WA dhufish and pink snapper. A growing catch-and-release fishery targets large, deep-water spawning aggregations (gatherings) of samson fish near Perth and this may have implications for sustainable management. Samson fish aggregations will also be used to examine a range of methods for studying and monitoring aggregations that may be applied to other aggregating species. Past information has been scarce, largely due to the fact that many aggregations are of short duration and occur in remote and deep locations, making them logistically and economically difficult to study. The high catch rates and large size of fish that may arise from aggregation fishing can also provide false indications about the health of the fishery. The area west of Rottnest Island where the samson fish aggregate appears to be an important aggregating site for other fish including mulloway, grey-banded cod, trevally and snapper. Little is known about the aggregations of these species, although they are increasingly vulnerable to fishing pressure as recreational fishing effort expands offshore. A current investigation of the biology of trevally and mulloway in the Perth region by Murdoch University will also be enhanced by additional information about the aggregating characteristics of these species in adjacent oceanic waters. The project will: * Identify species that aggregate to spawn within the West Coast Bio-region and to describe (e.g. location, size, timing, nature) the aggregations of key demersal species such as snapper and dhufish. * Investigate the biology, ecology and fishery for samson fish with emphasis on the sports fishery targeting deep water spawning aggregations west of Rottnest Island. * Establish methods and protocols for monitoring fish aggregations. * Review relevant information and provide advice on the impact of aggregation fishing and the management of aggregating fish species in WA (with specific advice for key species within the West Coast Bioregion). Time: Due for completion 2007

Data represent a comparison of herding techniques commonly used by natural resource agencies and the public to increase removal or harvest of invasive carp (i.e., Silver Carp) from U.S. waterways. Sites on lower Perche Creek, Columbia, MO (2018 August 9th to 2018 October 26th) were contained using block nets and treated with one of five herding techniques: (1) method commonly used by commercial fishers in the U.S. (“commercial technique”), (2) pulsed-DC electrofishing (“electric technique”), (3) broadband sound administered with underwater speakers (“sound technique”), (4) both sound and electric in combination (“combination technique”), and (5) solely the boat with no added stimulus (“control”). Sites were administered with repeated 20 min runs (x3) in a slow bank to bank pattern downstream through a 4.5 m opening along the shoreline. Herding effectiveness was quantified by run category (cumulative for second and third runs) using ARIS (Adaptive Resolution Imaging Sonar; Sound Metrics, Bellevue, WA). Fish composition of the study reach was sampled using a suite of community sampling gears downstream of a herding treatment on 2018 November 7th. Prior to the study, a playback sound recording was created for the sound technique and combination technique. In addition, sound pressure level (decible re 1 μPa) emitted from each technique were measured on Perche Creek.

Data represent a comparison of herding techniques commonly used by natural resource agencies and the public to increase removal or harvest of invasive carp (i.e., Silver Carp) from U.S. waterways. Sites on lower Perche Creek, Columbia, MO (2018 August 9th to 2018 October 26th) were contained using block nets and treated with one of five herding techniques: (1) method commonly used by commercial fishers in the U.S. (“commercial technique”), (2) pulsed-DC electrofishing (“electric technique”), (3) broadband sound administered with underwater speakers (“sound technique”), (4) both sound and electric in combination (“combination technique”), and (5) solely the boat with no added stimulus (“control”). Sites were administered with repeated 20 min runs (x3) in a slow bank to bank pattern downstream through a 4.5 m opening along the shoreline. Herding effectiveness was quantified by run category (cumulative for second and third runs) using ARIS (Adaptive Resolution Imaging Sonar; Sound Metrics, Bellevue, WA). Fish composition of the study reach was sampled using a suite of community sampling gears downstream of a herding treatment on 2018 November 7th. Prior to the study, a playback sound recording was created for the sound technique and combination technique. In addition, sound pressure level (decible re 1 μPa) emitted from each technique were measured on Perche Creek.

Underwater video technology was used to observe marine mammal interactions (seals, in particular Australian fur seals) within midwater trawl nets. A camera system was placed inside the trawl net in the vicinity of a Seal Excluder Device, designed to prevent megafauna from entering the codend of the net and provide an escape point. Digital video data was described according to a range of operational, catch and interaction fields, and made time-specific to each trawl shot. Data is linked to commercial catch and effort data providing operational, environmental and catch information.