I am currently pursuing the Ph.D. degree with the College of Computer Science and Technology, National University of Defense Technology, under the supervisor of Prof. Wenjing Yang and Prof. Long Lan. I also collaborate with Prof. [Jing Zhang] and Prof. [Bo Du] of Wuhan University in the remote sensing field. Any questions are welcome to contact me by wfx23@nudt.edu.cn.

MLLMs: I have carried out a series of works under the guidance of Professors [Zhiyuan Liu] and [Maosong Sun] from the THUNLP lab at Tsinghua University.

AI4S: my research has been conducted under the supervision of Professor [Lei Bai] at Shanghai AI Lab.

Research Highlights

MLLM Training: My research focuses on the training of advanced multimodal models, with a particular emphasis on super-high-resolution models such as GeoLLaVA-8k (NeurlPS 25 Spotlight).
MLLM Evaluation: I contribute to multiple research initiatives aimed at enhancing the evaluation processes for large-scale models. This includes projects such as XLRS-Bench (CVPR 25 Highlight), OmniEarth-Bench(Arxiv), SFE(NeurlPS 25).
Visual Foundation Model: My research also focuses on the visual foundation models for remote sensing, such as SelectiveMAE (ICCV 25) and RoMA (NeurlPS 25).
Prompt Tuning of VLMs: My work also includes the development of prompt tuning techniques for vision-language multimodal models, such as LoL (AAAI 24) and RS-tuning (TGRS).

🔥 News

2025.09: 🎉Two papers are accepted by NeurlPS 2025, and GeoLLaVA-8K has been selected as Spotlight.

📝 Publications

Conference

CVPR 2025 Highlight

Fengxiang Wang, H. Wang, Z. Guo, D. Wang, Y. Wang,… Zhiyuan Liu & Maosong Sun. XLRS-bench: Could your multimodal LLMs understand extremely large ultra-high-resolution remote sensing imagery? CVPR, 2025. （CCF-A类会议, CVPR Highlight，前3%，共384篇）
[arXiv] [Dataset] [Code]

NeurIPS 2025 Spotlight

Fengxiang Wang, M. Chen, Y. Li, D. Wang, H. Wang, … Bo Du & Jing Zhang. GeoLLaVA-8K: Scaling Remote-Sensing Multimodal Large Language Models to 8K Resolution. NeurIPS, 2025. （CCF-A 类会议, NeurIPS Spotlight, 前3.1%, 共688篇）
[Github] [arXiv] [Dataset]

NeurIPS 2025

Fengxiang Wang, H. Wang, Y. Wang, D. Wang, M. Chen, … & J. Zhang. RoMA: Scaling up mamba-based foundation models for remote sensing. NeurIPS, 2025. （CCF-A 类会议）
[arXiv] [Code]

ICCV 2025

Fengxiang Wang, H. Wang, D. Wang, Z. Guo, Z. Zhong, … & J. Zhang. Harnessing massive satellite imagery with efficient masked image modeling. ICCV, 2025. (CCF-A类会议)
[arXiv] [Code]

AAAI 2024

Fengxiang Wang, W. Huang, S. Yang… & L. Lan. Learning to learn better visual prompts. AAAI, 2024. (CCF-A 类会议)
[Paper]

arXiv 2025

Fengxiang Wang, M. Chen, X. He, Y. Zhang, F. Liu, … & Lei Bai. OmniEarth-Bench: Towards Holistic Evaluation of Earth’s Six Spheres and Cross-Spheres Interactions with Multimodal Observational Earth Data. Arxiv, 2025.
[arXiv] [Code]

Journal

TGRS 2024

L. Lan, Fengxiang Wang*, X. Zheng, Z. Wang, & X. Liu. Efficient prompt tuning of large vision-language model for fine-grained ship classification. IEEE TGRS, 2024. （SCI, 中科院一区Top，IF=8.6）
[arXiv]

💻 Work experience