[21p-P12-1~54] 17 Nanocarbon Technology(Poster)

[21p-P12-4] Re-learning super-resolution method for carbon nanotube images using SwinIR

〇Yoshiki Kakamu¹, Takahiro Maruyama¹, Kazuhiro Hotta¹ (1.Meijo Univ.)

Keywords:carbon nanotube, Deep Learning, Super-resolution

This paper presents super-resolution of carbon nanotube (CNT) images using Deep Learning. The problem of blurred images often arises in transmission electron microscopy, a technique for CNT evaluation. To solve the problem, we perform super-resolution of CNT images based on SwinIR. In addition, we propose a re-learning method to perform super-resolution with high accuracy even with a small amount of data. Experimental results show that both super-resolution and noise reduction can be achieved by using the re-learning method.