Underwater Sea Cucumber Detection Using Consumer-Grade Amphibious UAV and Deep-Learning Based Computer Vision

Holothurians (commonly known as sea cucumber) have economic and ecological value, as they are considered fishery resources, nutrient regulators, bioturbators, and indicator species in coral reef ecosystems. Therefore, continuous monitoring of Holothurians provides insights into ecosystem status and decision-making for ecological conservation and resource management. Traditional monitoring techniques are labor-intensive, constrained by the hardware, limited by obsolete algorithms, lack professional ecological information, and only cover a limited spatial scale. Based on a field survey conducted in Koh Tao, Thailand, a novel Holothurians monitoring system was established by combining a self-designed consumer-grade amphibious Unmanned Aerial Vehicle (AUAV) with a Mamba-based Super-Resolution Reconstruction (SRR) technique and an enhanced YOLOv10 instance segmentation model. Super-resolution (SR) images reconstructed by the MambaIR model were input into the proposed YOLOv10 SEA model for instance segmentation of Holothurians. Habitat maps and demographic parameters were measured based on the output segmented map. The results show that the proposed system can monitor Holothurians efficiently and accurately, providing useful ecological information to conservation communities and decision-makers.