Towards Automatic 2D Materials Detection Using YOLOv7

Two-dimensional (2D) materials have become a research focus in the field of nanotechnology. Because of their unique physical and chemical properties, these materials can be applied to various industries such as sensors, batteries, and display screens. However, identifying if optical microscope images contain 2D materials flakes is a tedious and time-consuming process. In this paper, we propose a deep learning-based object detection system using YOLOv7, to automatically search for 2D materials with few atomic layers (thickness in between 1 to 13 layers). To train the model, we have constructed Molybdenum Disulfide (MoS2) image dataset that is captured in 20x magnification and applied digital image processing and data augmentation techniques to increase the model performance. In addition, we developed a software pipeline that can be easily integrated with the optical microscope to detect 2D materials and store the result in the database. The result of the experiments shows that the trained model achieves high accuracy in detecting a few layers of MoS2.