Recently, the research An End-to-end Steel Surface Defect Detection Approach via Fusing Multiple Hierarchical Features by Professor Yan Yunhui's team from NEU’s School of Mechanical Engineering and Automation was selected as the“Popular Articles of 2020” of internationally well-known journal IEEE Transactions on Instrumentation and Measurement. This research was jointly completed by doctoral student He Yu, Associate Professor Song Kechen, and Professor Yan Yunhui from NEU and the team of Professor Qinggang Meng from Loughborough University.
Targeting real hot rolled steel plate surface defect detection, this study proposes an end-to-end defect detection method integrating multi-level features of convolution neural network, and compares this method with the current mainstream detection methods. Experiments show that this method has significant advantages in the precision and speed of detection and is of important value to improving the production quality of industrial products and promoting the R&D of intelligent detection technology in the manufacturing industry.
In recent years, NEU’s School of Mechanical Engineering and Automation has developed a series of scientific and effective supporting systems, which has guaranteed the cultivation of high-level research achievements and continuously enhanced the school's international academic influence.