ʻIKE Solutions founder Matthew Carnes has left his position as JIMAR PIFSC Fisheries Electronic Monitoring Associate in support the National Marine Atmospheric Administration to bring his talents to ʻIKE Solutions full time. During his time at RCUH, Matthew was in charge of maintaining 20 EM systems in the Hawaiʻi long line fisheries, as well as was the lead author on a technical memorandum demonstrating the efficacy of the technology in the region. Key results from NOAA Technical Memorandum NMFS-PIFSC-90 included the following:
Comparison of data collected by at-sea observers with post-cruise review of EM data indicate EM systems provide an additional means to accurately enumerate fish. A total of 89% of all catch enumerated by at-sea observers (retained and bycatch) were detected in EM data during video review. For retained fish only, EM reviewers located 98% of the fish enumerated by at-sea observers in the shallow-set fishery and 100% in the deep-set fishery. EM data also provided accurate enumeration over broad taxonomic groupings (e.g., tunas, billfishes, sea turtles) and for many economically valuable fish species. However, compared to at-sea observers, EM reviewers were not able to provide identifications to the species level for some species, including those subject to management implications, such as bigeye tuna and hardshell sea turtles. For bigeye tuna, there were significant differences (p < 0.001) between EM and at-sea observer enumerations. Sea turtle identifications were limited to the broader categories of hardshell or softshell sea turtle. Specific modifications to the current EM systems and catch handling are recommended in this technical memorandum to improve enumeration and identification to species that can be used for monitoring in the Hawaiʻi longline fisheries (e.g., identification of tunas, sharks, and sea turtles to species). Finally, this PIFSC pre-implementation project demonstrated that EM could be a cost-effective method to augment fisheries-dependent data collection. EM data review may be completed with a 76% reduction in the time needed to collect similar data by at-sea observers.
Matthew also pioneered using data collected from EM systems in active research. A seabird mitigation research project was conducted to evaluate how to best deploy tori lines, lines meant to prevent seabirds from being able to contact bait during setting operations, by using a stern mounted camera to record interactions behind the vessel for later review to determine contact and attempt rates. Using cameras helped make the research more affordable and able to be carried out during the seabird season, which only lasts 4-5 months.
Along with EM supported by human review, Matthew brings experience with machine learning tools to automate much of the tedious review needed for the successful implementation of EM in the region. With a combination of donations of images from the local fishery and open source images from fishnet.ai, Matthew was able to produce an ML model with about an 85% success rate to identify fish to species.
With years of experience in the field, Matthew is confident that this new chapter will be a positive endeavor. By coming to ʻIKE Solutions full time, the company will be able to become involved with more projects to help grow the company into what it was first conceived to be, a leader in providing technology to create a sustainable fishing future.