Seminar Series - Deep Inpainting, Deep Compressive Sensing, TEMImageNet and Autonomous TEM - Huolin Xin, Physics & Astronomy, University of California Irvine

Tuesday, May 4 at 2 p.m. via Zoom


Speaker: Professor Huolin Xin, Physics & Astronomy, University of California Irvine


Abstract: Driving a microscope is a multiscale on-the-fly computer vision (CV) problem. The key repeatable workflow is finding the sample at low magnifications and zoom in onto samples or regions of interest and record images at the desired resolutions. Semi-autonomy of this workflow has been achieved for single-particle protein and semiconductor IC device imaging. In these two applications, the feature of interest can be templated and easily located using traditional CV methods. However, the challenge for TEM imaging of a broader set of samples is that every single one of them looks different and the scales could be very different as well. In this talk, I will give an introduction to DeepEM Lab’s research on a series of deep-learning-based research toward enabling fully autonomous transmission electron microscopy. I will first talk about the building of a TEMImageNet and AtomSegNet—a training library and a suite of deep learning models for high-precision atom segmentation, localization, denoising, and super-resolution processing of atom-resolution STEM Images. Then, I will talk about mapping the mathematically ill-defined inverse problem in missing-wedge electron tomography to a CV inpainting problem. Finally, I will talk about the deep compressive sensing for super-compression of large electron microscopy timeseries.


Professor Huolin Xin   About Professor Huolin Xin: Huolin Xin is an associate professor at UC Irvine. He graduated from the Physics Department of Cornell University in 2011 and joined the University of California, Irvine in 2018. Prior to becoming a professor at UCI, he worked at Brookhaven National Laboratory as a scientific staff member and a principal investigator from 2013 to 2018. His research spans the areas from tomographic and atomic-resolution chemical imaging of battery and fuel cell materials to in situ environmental study of heterogeneous catalysts, and to the development of deep learning-enabled self-driving TEM. His research has resulted in more than 270 peer-reviewed publications (h-index 62 and citations 18,000). He received the MRS Oustanding Early Career Invetigator Award, MSA Burton Medal, DOE Early Career Award and the UCI Distinguished Early-Career Faculty for Research in 2020. He is the Chair of the largest international electron microscopy conference, Microscopy and Microanalysis, in 2020. His work on battery materials has been selected as the 2020, 2019 and 2014's Top-10 Scientific Achievements by Brookhaven Lab.