基于改进的YOLOv5的金属表面缺陷检测
Chuande Zhou1, Zhenyu Lu1, Zhongliang Lv2
1School of Mechanical and Power Engineering, Chongqing University of Science and Technology, Chongqing, 401331, China.
Scientific reports
|November 27, 2023
概括
本研究介绍了一种改进的YOLOv5s模型,用于检测小型金属表面缺陷,通过集成CSPlayer模块和全球注意力机制 (GAM) 来提高检测准确度和速度. 新型号在GC10-DET数据集上显著优于原来的YOLOv5s.
科学领域:
- 材料科学 材料科学 材料科学
- 计算机视觉 计算机视觉
- 人工智能的人工智能
背景情况:
- 由于复杂的纹理,金属生产中的表面缺陷难以检测,导致错误阳性或错过检测.
- 现有的小缺陷检测方法在准确性和速度方面扎,特别是在复杂的工业环境中.
研究的目的:
- 开发一种强大而高效的深度学习模型,用于检测金属表面上的小缺陷.
- 通过结合新的架构模块和注意力机制来提高YOLOv5s模型的性能.
主要方法:
- 通过将C3模块替换为CSPlayer模块以提高灵活性和适应性,YOLOv5s模型得到了增强.
- 整合了全球注意力机制 (GAM),并构建了一个通用增材模型以优化检测速度和准确性.
- 所有维度的注意力权重均为有效处理.
主要成果:
- 增强的YOLOv5s模型在GC10-DET增强数据集上表现出高于原始YOLOv5s的性能.
- 精度提高了5.3%,mAP@0.5提高了1.4%,mAP@0.5:0.95提高了1.7%.
- 改进的模型还实现了更高的推理速度,表明效率提高.
结论:
- 拟议的模型通过提高准确性和速度,有效地解决了金属材料中小缺陷检测的挑战.
- 集成CSPlayer模块和全球注意力机制 (GAM) 为提高工业应用中的对象检测任务提供了一个有希望的方法.
相关概念视频
01:00Detection of Gross Error: The Q Test
When one or more data points appear far from the rest of the data, there is a need to determine whether they are outliers and whether they should be eliminated from the data set to ensure an accurate representation of the measured value. In many cases, outliers arise from gross errors (or human errors) and do not accurately reflect the underlying phenomenon. In some cases, however, these apparent outliers reflect true phenomenological differences. In these cases, we can use statistical methods...
01:25Imperfections in Crystal Structure: Point, Line and Plane Defects
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...