Semantic segmentation benchmark dataset for coastal ecosystem monitoring based on unmanned aerial

Monitoring coastal ecosystems is an important task in maintaining ecological balance and sustainable development. The application of semantic segmentation technology in remote sensing images provides an effective means for precise monitoring of coastal ecosystems. However, there is still a challenge in this field, which is the lack of publicly available semantic segmentation benchmark datasets specifically for coastal ecosystems. Due to the combined influence of natural environment and human factors, the external morphology of the coastal zone changes rapidly. Currently, satellite remote sensing and conventional monitoring methods such as manual measurement and ship borne operations are unable to meet the requirements of real-time observation. Therefore, this article first uses drones to capture, collect, and annotate typical ecological communities in the coastal zone, and constructs a benchmark dataset for coastal ecosystems called OUC-UAV-SEG. Subsequently, quantitative analysis was conducted on OUC-UAV-SEG using statistical methods, and in-depth exploration was conducted on the challenges present in the dataset. Finally, evaluate OUC-UAV-SEG using typical visual semantic segmentation algorithms.