深度学习在工业表面缺陷检测领域的应用研究

Research on Application of Deep Learning in Field of Industrial Surface Defects Detection

深度卷积神经网络在处理自然图片时取得了非常好的效果,但鲜有针对工业应用领域的细分研究。本文探讨了深度学习模型在工业产品表面缺陷检测领域的应用。以Cp工业产品缺陷检测为着眼点,在设计检测方案时应用深度学习模型并辅助图像处理等相关技术,通过实验分析得到最佳应用模型。创新点在于提出了数据集信息密度这一概念,通过在多个数据集上的实验分析,论证了数据集信息密度对于深度学习模型的应用具有一定的评判价值。最后总结出一套具有一定普适性的深度学习模型在工业表面缺陷检测领域的应用指导方法。

Deep convolutional neural networks have achieved very good results when processing natural images,but there are few subdivision studies aimed at industrial applications.This article discusses the application of deep learning models in the field of surface defect detection of industrial products.Focusing on the defect detection of Cp industrial products,applying deep learning models and assisting related technologies such as image processing when designing the detection plan,and obtaining the best application model through experimental analysis.The innovation of this paper is to propose the concept of data set information density.Through experimental analysis on multiple data sets,it is demonstrated that the data set information density has a certain guiding value for the application of deep learning models.Finally,a set of guidance methods for the application of deep learning models with a certain universality in the field of industrial surface defect detection are summarized.