Beach-imaging derived beach wracks

This is a draft-dataset providing a time series of beach wrack coverage (COVR) at Cala Millor beach, derived from imagery captured by the SIRENA beach video-monitoring system. SIRENA's cameras acquire overlapping, hourly, oblique images at a resolution of 1280×960 pixels. A deep learning-based object detection and segmentation model identifies seagrass wracks in these images. Photogrammetric techniques then rectify the images to a terrain coordinate system, and seagrass wrack coverage is calculated as the area fraction per 5 m grid cell. To ensure data validity, a preliminary quality control step filters out detections landwards from the promenade seawall.本数据集为草案数据集，涵盖源自SIRENA海滩视频监测系统拍摄影像的卡拉米约尔海滩（Cala Millor Beach）海滩海草漂积覆盖度（beach wrack coverage, COVR）时间序列数据。SIRENA搭载的摄像头以1280×960像素的分辨率，每小时采集一组重叠倾斜影像。基于深度学习的目标检测与分割（object detection and segmentation）模型可从上述影像中识别海草漂积物。随后通过摄影测量技术（photogrammetric techniques）将影像校正至地形坐标系（terrain coordinate system），并以每5米网格单元的面积占比计算海滩海草漂积覆盖度。为保障数据有效性，本数据集设置了初步质量控制步骤，过滤掉滨海步道海堤向陆区域的检测结果。