On the 23rd of July, Dr. Ji Zhao gave a guest talk in our EORC talk seminar series. The lecture was titled “Deep Learning for Semantic Segmentation in Urban Remote Sensing: From Surface Water Mapping to Land Cover Classification“.

Here is the abstract of the presentation: Semantic segmentation plays a vital role in urban remote sensing for monitoring surface water dynamics and mapping land cover types. This talk introduces some deep learning-based methods developed for urban environments, including an unsupervised cross-sensor framework for large-scale water mapping, and a CNN–Mamba fusion network for land cover classification. These methods aim to improve segmentation accuracy, scalability, and generalization, providing solutions for environmental monitoring and geospatial analysis.