Automatic weld joint tracking method based on deep neural network

Abstract

The invention relates to an automatic weld joint tracking method based on a deep neural network. The automatic weld joint tracking method comprises a feature point detection network and a feature point tracking network. Feature point position information output by the feature point detection network is input into the feature point tracking network to realize automatic weld joint tracking. The feature point detection network is composed of a feature extraction module, an attention mechanism module, a priori box generation module and an identification positioning module. A weld joint feature point tracking network is composed of a feature extraction module, an attention mechanism module and a response output module. According to the method, automatic weld joint tracking is achieved through the neural network, and the weld joint feature point extraction network and the tracking network are designed according to weld joint image features; and by means of the method, the welding efficiencyand quality can be improved, and the adaptive capacity of an automatic weld joint tracking system in an actual complex welding environment is improved.

Description

technical field [0001] The invention belongs to the technical field of computer vision, relates to the field of deep learning and seam tracking, in particular to an automatic seam tracking method based on a deep neural network. Background technique [0002] Welding automation technology is widely used in industrial production due to its excellent work efficiency. It is an inevitable trend in the development of welding technology and an important means to achieve safer, more efficient and intelligent production. Real-time seam tracking is the key to welding automation. In today's society, with the rapid development of machining, automobile, construction and other manufacturing industries, metal welding has become an indispensable link in the production process. The quality of welding directly affects the quality and production efficiency of the entire product. With the gradual progress of social economy, industrial automation and science and technology, the demand for weldin...

AI Technical Summary

Problems solved by technology

[0003] During the welding process, due to the interference of strong arc light, spatter, smoke, etc., the collected images are seriously polluted by noise, and it is often difficult to accurately locate the feature points of the weld seam, which makes the welding seam tracking inaccurate and leads to welding deviation, which will Directly affect welding quality

Images