基于光度立体的复杂结构部件表面缺陷检测的数据集
1School of Life Sciences, Beijing University of Chinese Medicine, Beijing, 102488, China.
Scientific data
|February 16, 2025
概括
一种新的深度学习方法通过使用光度立体视觉和一种新型图像采集技术来改进金属表面的自动光学检查 (AOI). 这种方法提高了缺陷检测的准确性,减少了具有挑战性的非平面部件的错误.
科学领域:
- 计算机视觉 计算机视觉
- 机器学习 机器学习
- 材料科学 材料科学 材料科学
背景情况:
- 自动光学检查 (AOI) 对于工业质量控制至关重要.
- 传统的AOI面临着阴影,反射率和非平面表面的挑战,导致检测不准确.
研究的目的:
- 为金属表面开发一种新的缺陷检测技术,克服AOI限制.
- 为培训和验证金属表面缺陷检测的深度学习模型创建一个全面的数据集.
主要方法:
- 提出了一种结合光度立体视觉和深度学习的强光照明图像获取 (SIIA) 方法.
- 开发了泰勒系列频道混合器 (TSCM) 来从多角度照明中创建伪色图像.
- 在金属表面缺陷数据集 (MSDD) 上利用色彩随机化进行数据增强和验证的物体检测模型 (FCOS,YOLOv5,YOLOv8,RT-DETR).
主要成果:
- 在MSDD上获得了86.1%的平均平均精度 (mAP),超过了传统方法.
- 拟议的技术有效地处理了AOI固有的阴影和反射性问题.
- 该MSDD包括138,585个单通道和9,239个混合图像,涵盖八种缺陷类型.
结论:
- 新的深度学习和光度立体视觉方法显著提高了金属表面的自动视觉检查.
- 开发的MSDD为推进工业缺陷检测研究提供了宝贵的资源.
- 该方法为使用通用物体探测器进行端到端缺陷检测提供了强大的解决方案.
更多相关视频
相关概念视频
01:29Three-Dimensional Analysis of Strain
798
Three-dimensional strain analysis is crucial for understanding how materials deform under stress, particularly in elastic, homogeneous materials. This method employs principal stress axes to simplify complex stress states into more understandable forms. Subjected to stress, a small cubic element within a material either expands or contracts along these axes, transforming into a rectangular parallelepiped. This transformation effectively illustrates the material's deformation. The principal...
798
01:25Imperfections in Crystal Structure: Point, Line and Plane Defects
156
A perfect crystal, in theory, has a uniform structure with the same unit cell and lattice points throughout. However, any deviation from this periodic arrangement is known as an imperfection or defect. These defects can be categorized into three types: point, line, and plane defects.Point defects occur when there is a deviation from the ideal due to missing atoms, displaced atoms, or additional atoms. These imperfections might occur due to imperfect packing during crystallization or because of...
156