Electronic monitoring project improving data collection efficiency
A multiyear electronic monitoring collaboration between the Alaska Fisheries Science Center and NOAA's Fisheries Information System Program is improving data collection safety and efficiency onboard vessels with limited access for human observers.
The partnership was set up to automate video analysis for length measurement and species identification and integrate electronic monitoring data into overall catch accounting in North Pacific Fisheries. It marks the first time that data gathered from electronic monitoring collection has been successfully used for catch estimation. This marks an important milestone in integrating electronic technologies into the capture, analysis, and integration of fishery-dependent data.
NOAA said a camera chute system for measuring halibut bycatch during release from trawler catches was created as part of the project. This provides a known orientation of the fish and lighting for the cameras, which are triggered by sensors when each fish passes through. The images are later sorted by an algorithm by class of fish or, in cases of high uncertainty, flagged for manual review. Algorithms make calibrated measurements of each fish, which are compiled to estimate the total number and weight of halibut released.
During pilot testing of the system on a range of vessels, the chute was incrementally improved with such additions as vibration-resistant electronics and hardware built into an increasingly sturdy design, said NOAA. Software improvements allowed even poor-quality images to be processed and results to be provided soon after sorting was completed.
Because catch estimates can be made more quickly and the fish spend less time out of the water, camera chute systems can reduce bycatch mortality of high-value, prohibited fish species, such as halibut in the North Pacific.
NOAA stated as speed and accuracy of species identification improves, use of camera chutes can be expanded to more species and sampling situations. After two years of data, the fish identification algorithms now reliably identify the most commonly imaged species with about 94% accuracy. These same functionalities are also being developed for a stereo camera rail system being tested on longline vessels.
LATEST PRESS RELEASES
Future generations will need to produce more with less. Read more
Actions for investment and financing opportunities together with the day to day practicalities of ru... Read more
Lilja Alfreðsdóttir, the Minister of Education, Science and Culture in Iceland announced the winners... Read more
February 2018 Fareham UK - Mercator Media Ltd, the international, market-leading B2B marine media bu... Read more