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 SelectiveNet-Based Method for Defect Classification in Semiconductor Manufacturing

Deep Learning based Defect classification and detection in SEM images: A Mask R-CNN approach

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

DSCU-Net: MEMS Defect Detection Using Dense Skip-Connection U-Net

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

A Simple Confidence-Supervised Model for High-Resolution Defect Recognition

Addressing Class Imbalance and Data Limitations in Advanced Node Semiconductor Defect Inspection: A Generative Approach for SEM Images

SEMI-DiffusionInst: A Diffusion Model Based Approach for Semiconductor Defect Classification and Segmentation

Advancing SEM Based Nano-Scale Defect Analysis in Semiconductor Manufacturing for Advanced IC Nodes

A novel joint segmentation approach for wafer surface defect classification based on blended network structure

Saliency-Based Defect Detection in Industrial Images by Using Phase Spectrum

Interactive Defect Segmentation in X-Ray Images Based on Deep Learning

Image defect classification of surface mount technology welding based on the improved ResNet model

Inspection and Classification of Semiconductor Wafer Surface Defects Using CNN Deep Learning Networks

Optimal Feature Selection for Defect Classification in Semiconductor Wafers

Knowledge augmented broad learning system for computer vision based mixed-type defect detection in semiconductor manufacturing

Multi-Stage Process Diagnosis Networks in Semiconductor Manufacturing

SEMU-Net: A Segmentation-based Corrector for Fabrication Process Variations of Nanophotonics with Microscopic Images

An Evaluation of Continual Learning for Advanced Node Semiconductor Defect Inspection

Adversarial Defect Detection in Semiconductor Manufacturing Process

Intelligent Detection and Classification of Surface Defects on Cold-Rolled Galvanized Steel Strips Using a Data-Driven Faulty Model with Attention Mechanism