Surface Defect Detection and Root Cause Analysis

Tianchen Liu,Haisong Gu,Haoran Yu,Fan Zhu

doi:10.34257/gjsfrivol20is3pg1

Abstract

Artificial Intelligence has played an increasingly important role in surface defect detection in recent years. At the same time, there are many challenges using deep learning for this area, such as the detection accuracy, shortage of data and, lack of knowledge of root cause of defects. To solve the problem of data shortage, we propose a taxonomy method called DataonomyTM to extend a meta defect datasets with a small number of samples for training defect classifiers. For the accuracy, we apply two latest deep neural network(DNN) architectures, Inception v3 and fully convolutional networks (FCN) so as not only to classify whether there are defects but also to make a pixel-wise prediction to inference the areas of defects. For those detected defects, we combine DNN with traditional AI methods to find root causes of detected defects. We use a generalized multi-image matting algorithm to extract common defects automatically. We apply this technology to identify defects that stem from systematic errors in the surface operation. Experimental results have shown great capability and versatility of our proposed methods.

Highlights

Artificial Intelligence has played an increasingly important role in surface defect detection in recent years
To solve the problem of data shortage, we propose a taxonomy method called DataonomyTMto extend a meta defect datasets with a small number of samples for training defect classifiers
A common approach [6] to address this problem is to use transfer learning, in which a pre-trained model, such as VGG and Inception V3, is accuracy, we apply two latest deep neural network(DNN) chosen and retrained on the target dataset by 11 architectures, Inception v3 and fully convolutional networks keeping the model architecture and parameter weights

Summary

INTRODUCTION

Relationships between different datasets and extracting a Visual inspection is a common task across the industry, and across the world. Using artificial intelligence method[1] for asphalt pavement pothole detection based on least squares support vector machine and neural network with steerable filter-based feature extraction. Another example[2]was to use GLCM to extract texture features for surface quality detection for steel sheet. Both architectures have decent accuracies to find defects Another issue in the surface visual inspection field is that besides the basic defect detection tasks, few researches have dealt with root cause analysis for the detected image defects. 3) A generalized multi-image matting algorithm is applied to a root-cause analysis from surface defects The rest of the paper is structured as follows.

DATAONOMYTM

FRAMEWORK FOR IMAGE DEFECT DETECTION

EXPERIMENTS

Findings

CONCLUSION