A Deep‐Learning Framework for the Automated Recognition of Molecules in Scanning‐Probe‐Microscopy Images. (4th November 2022)

• Record Type: Journal Article
• Authors: Zhu, Zhiwen; Lu, Jiayi; Zheng, Fengru; Chen, Cheng; Lv, Yang; Jiang, Hao; Yan, Yuyi; Narita, Akimitsu; Müllen, Klaus; Wang, Xiao‐Ye; Sun, Qiang
• Is Part Of: Angewandte Chemie. Volume 134:Number 49(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-11-04
• Subjects: Computer Vision -- Machine Learning -- Mask R-CNN -- Molecular Recognitions -- Scanning Tunneling Microscopy; Chemistry -- Periodicals; 540
• DOI: 10.1002/ange.202213503
• Languages: English
• ISSNs: 0044-8249

Abstract

Abstract: Computer vision as a subcategory of deep learning tackles complex vision tasks by dealing with data of images. Molecular images with exceptionally high resolution have been achieved thanks to the development of techniques like scanning probe microscopy (SPM). However, extracting useful information from SPM image data requires careful analysis which heavily relies on human supervision. In this work, we develop a deep learning framework using an advanced computer vision algorithm, Mask R‐CNN, to address the challenge of molecule detection, classification and instance segmentation in binary molecular nanostructures. We employ the framework to determine two triangular‐shaped molecules of similar STM appearance. Our framework could accurately differentiate two molecules and label their positions. We foresee that the application of computer vision in SPM images will become an indispensable part in the field, accelerating data mining and the discovery of new materials. Abstract : We developed a deep‐learning framework to automatically analyse scanning‐probe‐microscopy data, which requires only a small number of image samples to achieve high accuracy. This work provides a strategy to accelerate data mining and the discovery of new materials.