微小辊印数据集：Steel-Tiny
Dataset of Tiny Roll Mark：Steel-Tiny
访问量：
数据集的介绍
Introduction of the dataset
微小辊印数据集Steel-Tiny是一个仅包含辊印的单一类型数据集。该数据集包含786张来自实际工业环境的缺陷图片，共计7315个辊印目标。辊印类缺陷的相对面积中位数为0.09%，如表1所示，符合Chen等人[1]对小目标的定义，即同一类别中，所有目标实例的相对面积的中位数在0.08%~0.58%之间。图1提供了局部微小辊印的放大细节。数据集格式符合PASCAL VOC标准，为研究微小辊印检测提供有力支持。
The Steel-Tiny dataset is a publicly available dataset specifically designed for the detection of tiny roll marks. It consists of 786 images captured from real industrial scenario, containing a total of 7,315 roll mark instances. The median relative area of roll mark defects is 0.09%, as shown in Table 1, which aligns with the definition of small objects provided by Chen et al. [1]. Specifically, within the same category, the median relative area of all target instances falls between 0.08% and 0.58%. Figure 1 provides a zoomed-in detail of a local tiny roll mark. The dataset follows the format of the PASCAL VOC standard, providing strong support for research on tiny roll mark detection.
所解决的问题
Problem solved
微小辊印数据集是业内首个小目标检测数据集，意图扩充小目标检测领域的研究对象，同时也对计算机视觉提出了相关挑战。
Our Steel-Tiny dataset is the first tiny object detection dataset in the industry. It aims to expand the research objects in the field of small object detection and also pose some challenges for computer vision research.
引用
Citation
如果您在使用数据库时引用我们的论文，我们将不胜感激:
We would appreciate it if you cite our works when using the database:
Qiwu Luo, Yangwen Chen, Jiaojiao Su, Chunhua Yang, Weihua Gui, O. Silvén, Li Liu. “Prior-Guided YOLOX for Tiny Roll Mark Detection on Strip Steel”. IEEE Sensors Journal, Mar. 2024, DOI: 10.1109/JSEN.2024.3374388
获取途径
Access
下载量：
[1] CHEN C, LIU M Y, TUZEL O, et al. R‑CNN for small object detection[C]//Proceeding of Asian Conference on Computer Vision. Cham: Springer, 2016: 214‑230.