Siegfried Dataset

该数据集名为Siegfried，包含了瑞士的历史地图数据，特别专注于葡萄园和铁路的分割任务。数据集被划分为224x224像素的图块，并包含了高达8000x8000像素的大规模扫描，以及一个框架遮罩，用于区分相关区域与非相关区域。具体规模上，铁路数据集包含5,872个训练图块，839个验证图块，以及1,679个测试图块；而葡萄园数据集则包含613个训练图块，87个验证图块，以及177个测试图块。该数据集的任务是少样本分割。

The dataset named Siegfried comprises historical cartographic data of Switzerland, with a particular focus on segmentation tasks for vineyards and railways. The dataset is divided into 224×224 pixel image tiles, and also includes large-scale scanned images up to 8000×8000 pixels in resolution, along with a frame mask to distinguish between relevant and irrelevant regions. In terms of specific scale, the railway segmentation subset contains 5,872 training tiles, 839 validation tiles, and 1,679 test tiles; while the vineyard segmentation subset includes 613 training tiles, 87 validation tiles, and 177 test tiles. The task of this dataset is few-shot segmentation.