2022 Volume 78 Issue 2 Pages I_703-I_708
In Yamaguchi Bay, where horseshoe crabs inhabit, a population survey is conducted every year to determine their population density. The problems of surveying larval horseshoe crab populations are that mudflats are difficult to walk on, require many people, the efficiency of finding individuals varies depending on the skill of the surveyor, and the visual survey area is unstable. The objective of this study was to develop a highly objective method of surveying the density of horseshoe crab larvae by detecting them in images of mudflat surfaces taken by a low-altitude UAV.
We investigated a method to detect horseshoe crabs on mudflats by taking images of the mudflat surface with a UAV at an altitude of 9 m, calculating the features of the trajectory, and performing machine learning, taking advantage of the fact that the trajectory of horseshoe crabs on mudflats has different characteristics from those of other benthic organisms and reflected light on the mudflats. Although it is possible to identify larval horseshoe crabs on mudflats, the results are not yet satisfactory, and further improvement of identification capability is an issue for the future.
