Xiaodong XU, Beijing University of Posts and Telecommunications, China
Xiaodong XU is currently a tenure professor of Beijing University of Posts and Telecommunications. His research area covers mobile networking theory, semantic communication system and trial systems. He accomplished the 4G and 5G trial networks with several awards, such as the first-class S&T award of China Institute of Communications. The 6G prototype he developed in 2021 was recognized as the “industry-leading” platform. He is an expert of IMT-2030 promotion group and CIC Fellow, IEEE Senior Member.

Yimu Ji, Nanjing University of Posts and Telecommunications, China
Yimu Ji is a Professor and Ph.D. supervisor at the School of Computer Science, School of Software, and School of Cyberspace Security, Nanjing University of Posts and Telecommunications, China. He is Deputy Director of the Institute of High Performance Computing and Big Data Processing and Executive Deputy Director of the Jiangsu Engineering Research Center for High Performance Computing and Intelligent Processing. His research covers high-performance computing, cloud computing, big data, artificial intelligence, the Internet of Things, and cybersecurity. His interests include parallel algorithms, data analytics, domain-specific large models, model optimization, and security for intelligent devices and networks. He has led or participated in national and provincial research projects and has published in journals and conferences spanning intelligent computing, information processing, and security. He was a visiting scholar at the University of Virginia in 2012 and Purdue University in 2015. His work connects computing infrastructure and intelligent algorithms with applications in industry.

 

Zhi Liu, School of Information Science and Engineering, Shandong University, China
Zhi Liu (Member, IEEE) received the Ph.D. degree from the Institute of Image Processing and Pattern Recognition, Shanghai Jiao Tong University in 2008. He is currently a Professor with the School of Information Science and Engineering, Shandong University. He is also the Head of the Intelligent Information Processing Group. His current research interests are intelligent computing, medical images in the neurosciences, multimodal human computer interaction, remote sensing image processing, content-based image retrieval, semantic modeling, and data mining, etc.

 

Ke Han, Institute of Forensic Science of China, China
Ke Han received the bachelor's degree in electronics and information technology from Shandong University, Jinan, China, in 2021, the master's degree in signal and information processing from Shandong University, Jinan, China, in 2024, and the doctor's degree in signal and information processing from Nanjing University of Posts and Telecommunications, Nanjing, China, in 2007. Since 2007, he has worked in Institute of Forensic Science of China, Beijing, China, and is currently a researcher with full senior professional title. His research interests concern artificial intelligence, image processing, pattern recognition, object detection, etc.

Submission Link (Enter the submission system and select Special Session 2 )

Overview:

Image segmentation methods based on artificial intelligence (AI) have made rapid development. However, under strongly-supervised learning of deep learning networks, it is necessary to do a lot of manual pixel-level feature annotation for images in some training datasets. The work of manually annotating image features is time-consuming, laborious, and susceptible to subjective influences. Images collected by different types of image acquisition devices have inconsistent feature distribution, which leads to the decline of generalization ability of deep learning networks under strongly-supervised learning. In order to solve the problems of manual pixel-level feature annotation and the decline of network generalization ability, image segmentation methods based on weakly-supervised learning, few-shot learning, transfer learning and source-target domain alignment, etc., have become one of the hot topics in current research. The purpose of these methods is to realize image segmentation under the condition of a small number of manual annotation or even no annotation of image features, and the segmentation effect is close to or even reach the effect of segmentation under the condition of strongly-supervised learning. In addition, deep learning networks should have certain generalization ability in image segmentation.
This special session considers approaches related to these aspects.

 

Related topics for this special session (but not limited to) :
1. Image segmentation under few-shot learning,
2. Image segmentation across source-target domains,
3. Image segmentation assisted by multimodal large models,
4. Image segmentation based on lightweight deep learning networks,
5. Image segmentation based on deep learning networks under prior constraints.