GaN Nanowire n-i-n Diode Enabled High-Performance UV Machine Vision System

Abstract

Machine vision as an essential component of artificial intelligence poses a significant influence on dimension measurement, quality control, autonomous driving, and so on. In this study, a high-performance ultraviolet (UV) imaging and detection system enabled by Gallium Nitride (GaN) nanowire (NW) n-i-n photodetector (PD) is presented. Based on supreme optoelectronic properties of the NW, including high responsivity of 5098 A/W, a low dark current of 4.88 pA and a photo-to-dark current ratio of 1223, machine vision system composed of a GaN NW array could achieve an accuracy of 96.21%. Furthermore, feasibility of artificial neural network (ANN) and convolutional neural network (CNN) in such a machine vision system is discussed, featuring dim and noisy environment. The visualization process shows that the superiority of CNN over ANN in image recognition is attributed to the capability of extracting spatial information and characteristics. The research results provide important insight into the development of both sensors and algorithms for machine vision systems based on GaN NW PD, inspiring further investigation into UV image detection and other areas of artificial intelligence.