Speaker: Zenghui Zhang

Affiliation: School of Electronics, Information, and Electrical Engineering (SIEEE) at Shanghai Jiao Tong University

Academic title: Professor

Abstract:

Deep learning techniques have demonstrated remarkable success in the field of remote sensing image classification. However, it is essential to acknowledge that deep learning models are highly reliant on the domain of their training data, leading to a significant decline in performance when applied to data from different domains. In this presentation, I will highlight recent research conducted by our laboratory addressing this critical issue. Specifically, I will introduce a novel 3D mask-MAE approach designed to simultaneously learn SAR and optical image features for classification. Additionally, I will discuss a self-supervised SAR classification method capable of extracting semantic features from geographically matched optical images, thereby enhancing SAR-based classification. The presentation will also address the challenges encountered and outline future research directions in this field.

Biography:

Zenghui Zhang has been working at the School of Electronics, Information, and Electrical Engineering (SIEEE) at Shanghai Jiao Tong University since 2013, where he currently is a professor. He serves as the Dean's Assistant at SEIEE and as the Associate Director of the Institute of Sensing and Navigation. His research interests include radar signal processing, SAR image interpretation, and remote sensing information processing. In these fields, he has published more than 100 journal and conference papers. Professor Zhang serves as an Associate Editor for Frontier in Remote Sensing and is an Editorial Board Member for Journal of Radars.