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).

GH-QFL: Enhancing Industrial Defect Detection Through Hard Example Mining

Coarse-to-Fine Few-Shot Defect Recognition with Dynamic Weighting and Joint Metric

Augmented Hybrid Learning for Visual Defect Inspection in Real-World Hydrogen Storage Manufacturing Scenarios

Few-shot defect detection using feature enhancement and image generation for manufacturing quality inspection

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

Human–machine knowledge hybrid augmentation method for surface defect detection based few-data learning

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

Cross-domain few-shot defect recognition for metal surfaces

Human-machine knowledge hybrid augmentation method for surface defect detection based few-data learning

Deep Learning for Just-In-Time Defect Prediction

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

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

HyperDefect-YOLO: Enhance YOLO with HyperGraph Computation for Industrial Defect Detection

A deep learning model for steel surface defect detection

Evaluating Generalization, Bias, and Fairness in Deep Learning for Metal Surface Defect Detection: A Comparative Study

Surface Defect Detection Algorithm Based on Feature-Enhanced YOLO

A real-time anchor-free defect detector with global and local feature enhancement for surface defect detection

An End-to-End Steel Surface Defect Detection Approach via Fusing Multiple Hierarchical Features

A Lightweight Deep Learning Network with an Optimized Attention Module for Aluminum Surface Defect Detection

TL-SDD: A Transfer Learning-Based Method for Surface Defect Detection with Few Samples

Yolo-sd: simulated feature fusion for few-shot industrial defect detection based on YOLOv8 and stable diffusion