Abstract:Visual surface defect inspection provides an important tool for product quality assessment in a wide range of industrial applications. In recent years, diverse convolutional neural network (CNN) models have been devel-oped for the high-accuracy image classification of natural objects. However, it remains a challenging task to develop deep-learning-based approaches for surface defect inspection of industrial products. The main challenge is that surface defects of industrial products occur in low probability, therefore it is impractical to build a large image dataset containing different types of defects to train/fine-tune deep CNN models to achieve satisfactory performance when applied to industrial inspection tasks. To overcome the above-mentioned problem, we build a light-weight defect classification model based on the pre-trained SqueezeNet architecture and present three effective techniques to achieve high-accuracy defect classification using very few collected defective training samples. First, we experimentally demonstrate that it is reasonable to freeze shallower convolutional layers and only fine-tune deeper layers in the last convolutional stage when there are not enough training samples. Second, we integrate batch-size independent Group Normalization (GN) into SqueezeNet to compute stable statistics based on a limited number of defective samples for training the classification model. Third, we add an auxiliary task to perform the similarity comparison of two defective samples during the training phase, significantly increasing the size of the training data and generating more distinctive defect-specific representations. The experimental results on the NEU-CLS dataset and USB-FS dataset verify the effectiveness of the proposed method. Our method can obtain accurate defect classification results when very few defective training samples are provided, i.e., about 97.69% and 82.92% average classification accuracy on the NEU-CLS and USB-FS datasets, respectively, under the 5-shot setting (5 images per defect category).

Tire Defect Detection Using Fully Convolutional Network.

Road Defect Detection from On-Board Cameras with Scarce and Cross-Domain Data

Automated Detection of Defects with Low Semantic Information in X-ray Images Based on Deep Learning

Tire Defect Detection Based on Low and High-level Feature Fusion

A customised ConvNeXt-SCC network: integrating improved principal component analysis with ConvNeXt to enhance tire crown defect detection

Visual Defect Inspection of Metal Part Surface via Deformable Convolution and Concatenate Feature Pyramid Neural Networks

Defect Detection in Tire X-Ray Images: Conventional Methods Meet Deep Structures

Tire tread detection based on fusion of spatial and frequency domain features

TD-YOLOA: An Efficient YOLO Network With Attention Mechanism for Tire Defect Detection

HLU 2 -Net: A Residual U-Structure Embedded U-Net With Hybrid Loss for Tire Defect Inspection

A Deep Learning-Based Surface Defect Inspection System Using Multiscale and Channel-Compressed Features

An Effective Industrial Defect Classification Method under the Few-Shot Setting Via Two-Stream Training

Detection and Segmentation of Manufacturing Defects with Convolutional Neural Networks and Transfer Learning

Structural Crack Detection Using Deep Learning–based Fully Convolutional Networks

Research on X-ray nondestructive defect detection method of tire based on dynamic Snake Convolution YOLO model

Unsupervised Learning with Generative Adversarial Network for Automatic Tire Defect Detection from X-ray Images

Spatio-Temporal 3-D Residual Networks for Simultaneous Detection and Depth Estimation of CFRP Subsurface Defects in Lock-In Thermography

An Efficient Network for Surface Defect Detection

Approaches for Accuracy Improvement of the X-ray Image Defect Detection of Automobile Casting Aluminum Parts Based on Deep Learning

Deep continuous convolutional networks for fault diagnosis

A transfer convolutional neural network for fault diagnosis based on ResNet-50