Labeled Images of Sand and Coins v2

179 images of sand and coins, and corresponding label image masks and associated files. For the purposes of training a model for image segmentation. Contents: 1. images.zip 2. masks.zip 3. rotimages4makesense.py images.zip contains a folder named 'images', itself containing 179 images. Imagery exif metadata have been modified to remove any 'auto-rotation' parameters, using the provided script `rotimages4makesense.py` masks.zip contains a folder named 'masks', itself containing 179 label images. Label images are white (value=255) where there is a coin, otherwise black (value=0) All images are 8-bit JPEG format and were acquired by the author. All label images were made by the author using the online program makesense.ai, downloading in VGG JSON format, then rasterizing the contents.本数据集包含179张沙与硬币的图像，以及对应的标签图像掩码与相关辅助文件，可用于图像分割模型的训练。 数据集组成详情如下： 1. images.zip：该压缩包内含名为`images`的文件夹，其中存储179张图像。为移除图像EXIF元数据中的自动旋转参数，作者通过配套脚本`rotimages4makesense.py`修改了所有图像的元数据。 2. masks.zip：该压缩包内含名为`masks`的文件夹，其中存储179张标签图像。标签图像中，硬币覆盖区域的像素值为255（白色），其余区域像素值为0（黑色）。 所有原始图像均为8位JPEG格式，由作者自行采集。所有标签图像均由作者通过在线工具makesense.ai制作：先以VGG JSON格式导出标注内容，再将标注栅格化得到最终的标签掩码。