MetricBased MetaLearning for CrossDomain FewShot Identification of Welding Defect

Abstract

With the development of deep learning and information technologies, intelligent welding systems have been further developed, which achieve satisfactory identification of defective welds. However, the lack of labeled samples and complex working conditions can hinder the improvement of identification models. This paper explores a novel method based on metricbased metalearning for the classification of welding defects with crossdomain fewshot (CDFS) problems. First, an embedding module using convolutional neural network (CNN) is applied to perform feature extraction and generate prototypes. The embedding module only contains one input layer, multiple convolutions, maxpooling operators, and batch normalization layers, which has the advantages of low computational cost and high generalization of images. Then the prototypical module using a prototypical network (PN) is proposed to reduce the influence of domainshift caused by different materials or measurements using the representations in embedding space, which can improve the performance of fewshot welding defects identification. The proposed approach is verified on real welding defects under different welding conditions from the CameraWelds dataset. For the Kshot classification on different tasks, the proposed method achieves the highest average testing accuracy compared to the existing methods. The results show the proposed method outperforms the modelbased metalearning (MAML) and transferlearning method.