题    目：Learning to Rank for Remote Sensing Image Interpretation

主讲人：刘亦书 教授

单    位：华南师范大学

时    间：2024年8月30日 9:00

腾讯ID：983 891 099

摘    要：Originating in the information retrieval domain, learning to rank (L2R) is the cornerstone of modern search and recommender systems. We have introduced L2R to remote sensing image interpretation in recent years, and have been trying to use it to build a unified interpretation framework. In this talk, we give a brief overview of L2R, and present two L2R-based interpretation models, namely R4C for multi-label aerial image classification and eFreeNet for remote sensing object counting. We also sketch out the blueprint for carrying out other interpretation tasks. Finally, we analyze what advantages L2R has over non-rank methods, and discuss why.

简    介：刘亦书，华南师范大学地理科学学院教授，博士生导师。主要研究兴趣包括机器学习、计算机视觉和遥感图像自动解译。迄今在国际顶级期刊发表论文30余篇，主持和参与国家级科研项目10余项。