Danfeng Hong

Research Statement

AI generated · Oct 26, 2026

Danfeng Hong's indexed work centers on hyperspectral image classification, where graph-based and transformer-based architectures recur across his most-cited first-author contributions. His graph convolutional networks for hyperspectral classification (2020) and SpectralFormer, which reframes classification around transformers (2021), are the two most-cited entries in this line, while multimodal deep learning for remote-sensing imagery classification (2020) extends the same problem to fused data sources. A parallel thread addresses spectral variability in unmixing through an augmented linear mixing model (2018), and more recent work scales toward foundation models with SpectralGPT (2024). Beyond these, Hong contributed to UIU-Net for infrared small object detection (2022) and to a comprehensive review of deep learning in multimodal remote sensing data fusion (2022), both middle-author roles that broaden the portfolio beyond hyperspectral classification. He has also produced benchmark resources, including cross-city multimodal datasets for domain-adaptive semantic segmentation (2023) and multimodal benchmark datasets for land cover classification (2021), alongside an overview and toolbox for shallow-to-deep feature extraction in hyperspectral imagery (2020). No software, datasets, grants, or patents are indexed beyond these publications. The arc runs from classical unmixing and feature extraction toward transformer and foundation-model approaches, with benchmark datasets supporting reproducibility along the way.

Most Cited Works

Graph Convolutional Networks for Hyperspectral Image Classification

Danfeng Hong, Lianru Gao, Jing Tao Yao, Bing Zhang, Antonio Plaza, Jocelyn Chanussot

IEEE Transactions on Geoscience and Remote Sensing2020

1,717Citations

More Diverse Means Better: Multimodal Deep Learning Meets Remote-Sensing Imagery Classification

Danfeng Hong, Lianru Gao, Naoto Yokoya, Jing Tao Yao, Jocelyn Chanussot, Qian Du, Bing Zhang

IEEE Transactions on Geoscience and Remote Sensing

Research Areas

Remote-Sensing Image ClassificationAdvanced Image Fusion TechniquesAdvanced Neural Network ApplicationsRemote Sensing and Land UseSatellite Image Processing and Photogram

Affiliations & Education

Education

2015–2021
German Aerospace Center(DE)
Remote Sensing Technology Institute
2016–2019
Technical University of Munich(DE)
Dr.

Employment

2025–present· now
Southeast University(BD)
Professor

Loading ORCID activity...

Sources