The aim of this work was to assess the possibilities of automatically generating a dataset of dive behaviour of air-breathing predators, based on acoustic data from a monitoring survey and from commercial krill fishing operations. Our results documents that some form of automatic detection of diving predators in the data is feasible. A relatively low detection probability of our algorithms compared to the manual detections, suggest that there is significant room for improvement. Given the caveats of an imperfect methodology, the results document the possibilities to automatically extract, with a reasonable level of precision, data on the dive behaviour of air-breathing predators from the echo-sounder data.
Dr Thor Klevjer (Norway)