Intrinsic Dimension Estimation of Iudustrial Surface Defect Data and Its Impact on Classification

In today's era of rapid development of artificial intelligence, machine vision technology based on deep learning is widely used. In this paper, the intrinsic dimension estimation of industrial defect data and its impact on classification and generalization are studied based on machine vision in the detection of industrial defects. In this work, Maximum Likelihood Estimation (MLE) algorithm is used to estimate the intrinsic dimensions of three common industrial defect data, and to verify the influence of the intrinsic dimensions on sample complexity and classification.