欢迎加入我们的团队
我们的研究聚焦于多模态空间智能、场景的结构化与组合式理解,以及相关技术在遥感解译和水下环境理解中的应用。具体涉及的技术方向包括:
- MLLM 辅助的空间推理,例如目标导航、视觉语言导航和多模态推理。
- 场景图生成,例如遥感场景图生成和水下场景图生成。
- 变化检测、语义分割和实例分割。
王文彬
王文彬博士毕业于中国科学院计算技术研究所(ICT)智能信息处理重点实验室(IIP)视觉信息处理与学习研究组(VIPL)(VIPL-VSU),导师为陈熙霖教授和王瑞平教授。此前,他于南开大学(NKU)计算机科学与技术专业获得工学学士学位(2013-2017)。他曾获2024年中国图像图形学学会博士学位论文激励计划提名,入选湖北省楚天学者人才计划。他受邀担任 IEEE TPAMI / IEEE TMM / PR / ESWA 等期刊审稿人和 CVPR / ICCV / ECCV / NeurIPS 等会议审稿人,是CAAI(智能服务专业委员会)、CCF、CSIG 会员。
🔥 新闻
- 2026.08: 一篇关于遥感场景图生成的论文被 IEEE TGRS 接收。
- 2026.08: 一篇论文被 IEEE JSTARS 接收。
- 2026.07: 一篇论文被 IEEE TGRS 接收。
- 2026.05: 一篇论文被 IEEE TGRS 接收。
- 2025.12: 一篇论文被 IEEE TCSVT 接收。
- 2024.12: 一篇论文被 The Visual Computer 接收。
- 2023.05: 一篇关于层次化场景图生成(SGoI)的论文被 IJCV 接收。
- 2022.07: 一篇论文被《中国科学:信息科学》接收。
- 2021.07: 一篇关于场景图生成和图像描述的论文被 ICCV 2021 接收。
- 2020.07: 一篇关于层次化场景图生成的论文被 ECCV 2020 接收。
- 2019.03: 一篇关于视觉关系检测的论文被 CVPR 2019 接收。
📝 论文发表
TGRS 2026

Spatially Anisotropic Reasoning Network for Remote Sensing Scene Graph Generation
Wenbin Wang, Yiheng Chen, Hang Sun, Dong Ren, Yansheng Li
- 遥感图像通常包含多个功能子场景,而关系语义会受到目标间相对空间构型的显著影响。
- SARNet 提出各向异性椭圆影响传播(AEIP),用于建模以目标为中心的空间影响并抑制带有噪声的长距离交互;同时利用相对空间构型注意力(RSCA)增强具有几何判别力的关系表征。
- 代码发布: SARNet
IJCV 2023

Importance First: Generating Scene Graph of Human Interest
Wenbin Wang, Ruiping Wang, Shiguang Shan, Xilin Chen
- 人类在分析自然场景时,对场景内容的感知存在层次化偏好。
- 本文提出 SGoI(Scene Graph of Human Interest)生成框架,根据内容重要性组织并呈现场景图。
- 代码发布: TGIR
ICCV 2021

Topic Scene Graph Generation by Attention Distillation from Caption
Wenbin Wang, Ruiping Wang, Xilin Chen
- 场景图生成和图像描述可以被看作场景理解的两种形式。
- 本文提出了一个共享框架,同时生成场景图和图像描述。
- 代码发布: MMSceneGraph
:一个基于 MMDetection 的场景图生成工具箱,包含多数 SOTA 场景图生成方法。
TGRS 2026Wenbin Wang, Yiheng Chen, Hang Sun, Dong Ren, and Yansheng Li. Spatially Anisotropic Reasoning Network for Remote Sensing Scene Graph Generation. IEEE TGRS, vol. 64, Art. No. 4416521, 2026.JSTARS 2026Lu Wang, Chenxu Lou, Jing Wang, Junbo Yu, Lihua Zhao, Wenbin Wang, and Dong Ren. Scale-aware Fusion and Spatial-Frequency Collaborative Network for Remote Sensing Imagery Semantic Segmentation. IEEE JSTARS, vol. 19, pp. 28361-28378, 2026.TGRS 2026Hang Sun, Guanbo Feng, Bo Du, Yibing Zhan, Dong Ren, Wenbin Wang *, and Jun Wan. Heterogeneous Prompting Cross-Domain Modulation Network for Remote Sensing Image Dehazing. IEEE TGRS, vol. 64, Art. No. 5631618, 2026.TGRS 2026Hang Sun, Runzhou Li, Chenyang Wang, Lefei Zhang, Dong Ren, and Wenbin Wang *. Frequency-Domain Heterogeneous Rank-Entropy Bipolarization Network for Remote Sensing Change Detection. IEEE TGRS, vol. 64, Art. No. 5621517, 2026.JSTAR 2025Lu Wang, Chenyang Wang, Runzhou Li, Junbo Yu, Hang Sun, Wenbin Wang *, and Dong Ren. MDFANet: Multidimensional Feature Alignment Network for Self-Supervised Pretraining in Remote Sensing Change Detection. IEEE JSTAR, vol. 19, pp. 2355-2372, 2025.TCSVT 2025Hang Sun, Qingfei Zhong, Bo Du, Zhigang Tu, Jun Wan, Wenbin Wang *, Dong Ren. Bidirectional-Modulation Frequency-Heterogeneous Network for Remote Sensing Image Dehazing. IEEE TCSVT, vol. 35, no. 11, pp. 10649-10664, 2025.TVC 2025Tang Xu, Wenbin Wang *, ALin Zhong. HOIEdit: Human-Object Interaction Editing with Text-to-Image Diffusion Model. The Visual Computer, vol. 41, pp. 6187-6199, 2025.ECCV 2024Yansheng Li, Tingzhu Wang, Kang Wu, Linlin Wang, Xin Guo, Wenbin Wang. Fine-Grained Scene Graph Generation via Sample-Level Bias Prediction. In ECCV, vol. 15084, pp. 18-35, 2024.[pdf] [DOI] [arXiv] [supp]
TPAMI 2024Yansheng Li, Linlin Wang, Tingzhu Wang, Xue Yang, Junwei Luo, Qi Wang, Youming Deng, Wenbin Wang et al. STAR: A First-Ever Dataset and A Large-Scale Benchmark for Scene Graph Generation in Large-Size Satellite Imagery. TPAMI, vol. 47, no. 3, pp. 1832-1849, 2024.[project]
IJCV 2023Wenbin Wang, Ruiping Wang, Shiguang Shan, Xilin Chen. Importance First: Generating Scene Graph of Human Interest. IJCV, vol. 131, no. 10, pp. 2489–2515, 2023.SCIENTIA SINICA Informationis 2022Wenbin Wang, Ruiping Wang, Xilin Chen. Balanced Scene Graph Generation Assisted by Additional Biased Predictor. Scientia Sinica Informationis, vol. 52, no. 11, pp. 2075-2092, 2022.[pdf]
ICCV 2021Wenbin Wang, Ruiping Wang, Xilin Chen. Topic Scene Graph Generation by Attention Distillation from Caption. In ICCV, pp. 15900-15910, 2021.[DOI] [pdf]
ECCV 2020Wenbin Wang, Ruiping Wang, Shiguang Shan, Xilin Chen. Sketching Image Gist: Human-Mimetic Hierarchical Scene Graph Generation. In ECCV, vol. 12358, pp. 222–239, 2020.[pdf] [data] [slides] [video] [arXiv] [proc.] [supp]
CVPR 2019Wenbin Wang, Ruiping Wang, Shiguang Shan, Xilin Chen. Exploring Context and Visual Pattern of Relationship for Scene Graph Generation. In CVPR, pp. 8188–8197, Long Beach, CA, US, 2019.[DOI] [pdf] [proc.]