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).

A graph-based two-stage classification network for mobile screen defect inspection

Few-Shot Classification of Screen Defects with Class-Agnostic Mask and Context-Based Classifier.

Grading of Diabetic Retinopathy Images Based on Graph Neural Network.

Surface Defect Detection of Mobile Phone Screen Based on Generative Adversarial Networks

Scale insensitive and focus driven mobile screen defect detection in industry

Context-Enhanced Network with Spatial-Aware Graph for Smartphone Screen Defect Detection

Open-Set Recognition of Screen Defects with Negative-Guided Augmented Prototype Generator and Open Feature Generation

Two-Stream Network One-Class Classification Model for Defect Inspections

Intra-Class Structure Aware Networks For Screen Defect Detection

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

Small-scale defect detection in industrial environment based on lightweight deep learning network

A new graph-based semi-supervised method for surface defect classification

Real Time Detection of Surface Defects with Inception-Based MobileNet-SSD Detection Network

A Novel Feature‐based Network with Sequential Information for Textile Defect Detection

A Small-scale Object Detection Method for LCD Defects Based on Improved YOLOv8

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

An Efficient End-to-End Multitask Network Architecture for Defect Inspection

Soft matching network with application to defect inspection

An Efficient Network for Surface Defect Detection

Mobile Phone Surface Defect Detection Based on Improved Faster R-CNN

FDSNeT: an Accurate Real-Time Surface Defect Segmentation Network.