Use of Video Cameras and AI to Monitor Recreational Fishing Vessels

The Fish and Wildlife Research Institute (FWRI) has a longstanding interest in improving recreational fishing estimates in Florida. This concerted endeavor complements federal fishing estimates and regulations. Essential data on these fisheries (e.g., number of trips, number of fish caught, released or harvested, time spent on the water, and locations fished) help fisheries management groups better understand the health of our stocks and adjust regulations as needed to support both fish and anglers.

FWRI’s Fisheries-Dependent Monitoring program was designed to collect data from both commercial and recreational fisheries in Florida. These data feed directly into stock assessments, are used to monitor fishing quotas and harvest caps, and further characterize these important economic industries for the state.  This team continues to explore the best methods for collecting these data, most recently through the implementation of camera monitoring systems on Florida’s Atlantic coast now led by Genine McClair. Camera monitoring has emerged as a globally utilized tool for observing both terrestrial and marine habitats. As the technology advances so do the possibilities for monitoring. These systems’ primary benefit is their ability to expand monitoring coverage beyond human sampling capabilities. Additionally, parameters such as spatial position, location, time of day and weather limitations are all made consistent with camera monitoring. In addition to improved consistency, cameras expand monitoring coverage. A camera can be present at all times which allows for a more precise census count compared to an extrapolated estimate, which is highly desired by decision makers. Camera surveys also reduce labor costs associated with monitoring.

Commercial fishing is well monitored and better understood due to registration and reporting requirements. McClair notes that commercial fisheries require much more stringent registration and licensing, which provide us with an accurate number of fishers and vessels participating in the fishing effort each year. Commercial fishers are also required to report what they harvested on each trip.  On the other hand, recreational fishing is more challenging to monitor. Not only are there more anglers and more trips taken annually, but there is also no mandated reporting of catch. It is important to gain an understanding of the recreational fishery to responsibly manage it.

The primary way we gain information on the recreational fishing effort is through FWC’s State Reef Fish Survey (SRFS) and NOAA’s Marine Recreational Information Program’s (MRIP) Fishing Effort Survey (FES). Both surveys mail out a questionnaire to gather specific information about locations fished, trips taken, and more. These survey methods are a widely accepted approach, but their efficacy depends on good response rates from anglers as well as focused and accurate reports. Many anglers opt out of the surveys all together, and variation in results is common due to the large number of individuals reporting data and the need for them to recall the information. McClair recently took lead on the camera project and says, “this effort is designed to test the technology’s ability to provide an estimate of the fishing effort that could be compared to the estimates produced through the mail surveys. The main benefit of the camera project is it provides a direct count of trips rather than relying on human recall.”

The project began with three cameras deployed at navigational inlets on Florida’s Atlantic coast. The cameras were placed in a manner that they would record boats passing through the inlet on their way to and from a day on the water. FWRI’s Tiffanie Cross has been working to develop site relations and install the camera systems for years in preparation for this project. First, suitable locations must be determined based on their comparability to other locations in recreational fishing data. Then, suitable sites must be willing to host a camera, and electricity and internet must be made available. Even if all of that works, “speed bumps”, as Tiffanie calls them, still arise. Even at sites with ideal infrastructure, water intrusion, power outages and connectivity issues have the potential to halt the project. The camera positioned at Ponce Inlet was exposed two hurricanes that damaged the structure supporting the camera and prevented access for over a year. The team recently reacquired the camera but discovered that its housing was damaged and suffered water intrusion. Another camera at Fort Clinch was struck by lightning and is no longer in operation. This project is still in its pilot phase, so troubleshooting and developments are all part of the process to ensure an accurate, efficient and helpful system is produced.

Despite some obstacles along the way, the results so far have kept the team hopeful. After the cameras were operational, an algorithm was developed to review footage, and trained by members of the team to ensure that it records data in the way a human researcher would. The team reviewed over a year of footage and data collected beginning in 2021, either by checking the system’s work or by using a human surveyor in conjunction with the AI surveyor to compare results. The algorithm was programmed to exclude vessels such as sailboats, tugboats, commercial ships, and other boats that are unlikely to be fishing, as human surveyors would also exclude these boats. Upon review, the AI surveyor showed good agreement with the results a human surveyor would achieve and is ready for consistent use. McClair says, “our goal moving forward this year is to attain continuous boat count data during summer months; this data will help researchers generate estimates of total fishing efforts on the Atlantic coast.” The cameras, positioned at Mayport and Ponce Inlet, are also able to record data at all times of day and in any weather condition, which has the potential to provide us with important data from trips departing before dawn. “Since the cameras have been recording data, we have the ability to include these early morning trips in our data. Our current sampling using dockside observers do not capture this peak time, so we’re looking forward to having more coverage during this important timeframe,” says McClair.

The consistency of the data collected through AI monitoring is expected to enhance the accuracy of recreational fishing estimates. Implementing systems like this across the Atlantic coast would provide consistency to sampling and a more complete estimate of recreational fishing efforts. Anglers in Florida can contribute to these important monitoring efforts by accurately completing surveys, whether by mail or a dockside sampler, whenever possible. When purchasing a saltwater recreational fishing license, remember to add the reef fish angler designation if you plan to fish for any of the following species: mutton snapper, yellowtail snapper, hogfish, red snapper, vermilion snapper, gag grouper, red grouper, black grouper, greater amberjack, lesser amberjack, banded rudderfish, almaco jack or gray triggerfish. Always complete and return the SRFS mail survey if you receive one; this supports FWC’s ability to estimate recreational fishing catch and effort for reef fish. The South Atlantic Fisheries Management Council also has citizen science initiatives for fishers to participate in. These efforts collectively support responsible and informed fisheries management decisions to conserve this important resource.