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).

Industrial fault diagnosis based on few shot learning

A Few-Shot Machinery Fault Diagnosis Framework Based on Self-Supervised Signal Representation Learning

An intelligent fault diagnosis model based on deep neural network for few-shot fault diagnosis

Few-Shot Learning for Fault Diagnosis With a Dual Graph Neural Network

Few-Shot Learning Approaches for Fault Diagnosis Using Vibration Data: A Comprehensive Review

Cross-Category Mechanical Fault Diagnosis Based on Deep Few-Shot Learning

Few-shot fault diagnosis based on multi-scale graph convolution filtering for industry

Center Loss Guided Prototypical Networks for Unbalance Few-Shot Industrial Fault Diagnosis

A Novel Fault Diagnosis Method of High-Speed Train Based on Few-Shot Learning

Few-shot transfer learning for intelligent fault diagnosis of machine

Few-shot intelligent fault diagnosis based on an improved meta-relation network

A General and Efficient Fault Diagnosis Approach Using Hyperdimensional Computing Enhanced by Local Feature Extraction

Few-Shot Learning for Fault Diagnosis: Semi-Supervised Prototypical Network with Pseudo-Labels

Few-Shot Fault Diagnosis Based on an Attention-Weighted Relation Network

Meta-Learning Guided Few-Shot Learning Method for Gearbox Fault Diagnosis Under Limited Data Conditions

Domain Discrepancy-guided Contrastive Feature Learning for Few-shot Industrial Fault Diagnosis under Variable Working Conditions

Category knowledge-guided few-shot bearing fault diagnosis

Reweighted Regularized Prototypical Network for Few-Shot Fault Diagnosis

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

Fault Diagnosis for Rolling Bearings of a Freight Train under Limited Fault Data: Few-Shot Learning Method