Welding Defect Recognition Technology and Application Based on Convolutional Neural Network

In industrial manufacturing, surface defect detection is the key step to ensure product quality. Typically, surface defect inspection is conducted by trained workers to identify complex surface defects. However, due to the complex shape of the product to be inspected and the different degree of illumination, the manual inspection efficiency is low, which takes a long time, and is greatly affected by manual experience and subjective factors. This paper proposes a defect detection model based on convolutional neural network. This network introduces atrous convolution to replace standard convolution, which reduce the calculation amount of the model and improve the real-time performance of defect detection. Finally, our proposed model is demonstrated on a dataset of industrial surface defects and compared with the mainstream methods. Experimental results show that the detection effect of our network is better than other related methods, and less time consuming.