CISS 2023

Title: Target Detection and Tracking in SAR images

Abstract:Taking advantage of the all-day and all-weather observation, Synthetic Aperture Radar (SAR) techniques have been widely used to track moving ground targets from satellite platform. Deep learning based algorithms have been sucessfully used for target detection and tracking in SAR images and presented lots of advantages. When they are applied to spaceborne SAR images, many difficulties need to be considered, such as the characteristics of high density and small scale of targets on spaceborne SAR images, the huge calculation amount and volume of deep learning algorithm models, and it is difficult to deploy on spaceborne equipment to achieve real-time object detection. In this presentation, we aim to solve the above problems by detailing two real-time object detection methods based on SAR images: 1) improving the model structure to enhance the feature extraction and fusion capabilities and then adopting pruning strategies to achieve lightweight and real-time object detection, and 2) adopting a lightweight backbone network design and optimizing the loss function to improve the detection accuracy of lightweight models based on knowledge distillation.?

Bio: Shaohui Mei received the B.S. degree in electronics and information engineering and the Ph.D. degree in signal and information processing from Northwestern Polytechnical University, Xi’an, China, in 2005 and 2011, respectively. He was a Visiting Student with the University of Sydney, Camperdown, NSW, Australia, from October 2007 to October 2008. He is currently a Professor with the School of Electronics and Information, Northwestern Polytechnical University. Dr. Mei received the First prize of Natural Science Award of Shaanxi Province in 2022, the Excellent Doctoral Dissertation Award of Shaanxi Province in 2014, the Best Paper Award of the IEEE ISPACS in 2017. He is serving as the Associate Editor for IEEE TGRS and IEEE JSTARS, Guest Editor for Remote Sensing, and the Reviewer for more than 30 international famous academic journals. His research interests include hyperspectral remotesensing image processing and applications, intelligent signal and information acquisition and processing, video processing, and pattern recognition.