SWIR_volume_estimation

The dataset consists of grayscale images containing up to four circular objects (samples) per image. The dataset follows a structured naming convention, where filenames encode the thickness of the material or the depth of holes present in the samples. Each image file follows the naming pattern cc-cc-cc-cc_0001.jpeg, where cc represents a two-digit number specifying either the material thickness or the hole depth in millimeters. The four cc values correspond to specific positions in the image: the first value represents the top-left object, the second value represents the top-right object, the third value represents the bottom-right object, and the fourth value represents the bottom-left object. If no object is present in a specific position, XX is used instead of a numerical value. The final _0001 is an incrementing counter for multiple images of the same set of objects, increasing with each occurrence. The numbers in the filenames correspond to sample specifications listed in an accompanying dataset index. The handwritten notes provided define the measurements as follows: 21 corresponds to 0.58 mm, 22 corresponds to 1.20 mm, 23 corresponds to 1.58 mm, 24 corresponds to 2.17 mm, 25 corresponds to 2.57 mm, 26 corresponds to 3.22 mm, 27 corresponds to 3.58 mm, 28 corresponds to 4.16 mm, and 29 corresponds to 4.57 mm. Similarly, hole specifications include 00 corresponding to a 5 mm hole, 01 corresponding to a 4.5 mm hole, going lineary by 0.5 mm decrease to vanule 09 corresponding to a 0.5 mm hole, and 10 corresponding to no hole. Each image displays up to four dark circular shapes representing material samples, some of which may have central holes of varying diameters. The grayscale shading might indicate variations in depth, surface texture, or material density. This dataset can be used for image processing and segmentation tasks, optical analysis of material properties, and machine learning applications related to material classification. If any additional metadata is required, it can be extracted from the naming conventions.本数据集由灰度图像构成，单张图像最多可包含四个圆形样品。该数据集遵循结构化命名规范，文件名编码了样品的材料厚度或孔深信息。所有图像文件均遵循cc-cc-cc-cc_0001.jpeg的命名格式，其中cc为两位数字，单位为毫米，分别代表材料厚度或孔深。四个cc数值依次对应图像中四个特定位置的样品：第一个数值代表左上角样品，第二个代表右上角样品，第三个代表右下角样品，第四个代表左下角样品。若某一位置无样品，则以XX替代数值。文件名末尾的_0001为同一组样品的多张图像的递增计数器，随图像生成顺序依次递增。 文件名中的数字与配套数据集索引中列出的样品规格一一对应。手写标注定义了如下数值映射关系：21对应0.58毫米，22对应1.20毫米，23对应1.58毫米，24对应2.17毫米，25对应2.57毫米，26对应3.22毫米，27对应3.58毫米，28对应4.16毫米，29对应4.57毫米。同理，孔规格的映射关系如下：00对应直径5毫米的孔，01对应直径4.5毫米的孔，后续数值以每级0.5毫米的幅度线性递减，直至09对应直径0.5毫米的孔，10则代表无孔。 每张图像展示最多四个深色圆形图案，代表材料样品，其中部分样品可能带有直径各异的中心孔。图像的灰度明暗变化可反映样品的深度差异、表面纹理或材料密度。 本数据集可应用于图像处理与分割任务、材料特性的光学分析，以及与材料分类相关的机器学习场景。若需获取额外元数据，可通过命名规范直接提取。