Boni Hu

I am now a 4th-year PhD (Sep. 2021 - ) student at Northwestern Ploytechnical University, affiliated with the Pilot Intelligent Laboratory (PI-LAB) and advised by Prof. Shuhui Bu. Previously, I obtained my M.Eng (Sep. 2018 - Mar. 2021) in Vehicle Operation Engineering from Northwestern Ploytechnical University and B.Eng (Sep. 2014 - Jun. 2018) in Electronic and Information Engineering from Northwestern A&F University.

My current research mainly focuses on vision representation for remote sensing in place recognition and geolocalization, as well as its applications in UAV platforms. I am deeply interested in exploring elegant vision representations of the world, particularly in applications related to unmanned and medical devices. I am currently seeking relevant postdoctoral positions .

Email  /  CV  /  Google Scholar  /  Github  /  Twitter

profile photo

Recent News
  • [03/2024] One paper accepted to TGRS (IF=8.2) and the code is released.
  • [01/2024] One paper accepted to TGRS (IF=8.2).
  • [07/2023] One paper accepted to National Remote Sensing Bulletin.
  • [07/2023] One paper accepted to TGRS (IF=8.2).
  • [01/2021] One paper accepted to ICRA 2021.
  • [07/2020] One paper accepted to MICCAI 2020.

Seleted Publications
Place Recognition/Geolocalization for UAV Platforms

CurriculumLoc: Enhancing Cross-Domain Geolocalization Through Multistage Refinement
Boni Hu, Lin Chen, Runjian Chen, Shuhui Bu, Pengcheng Han, Haowei Li
IEEE Transactions on Geoscience and Remote Sensing, 2024
IEEE paper / code

Inspired by curriculum design, human learn general knowledge first and then delve into professional expertise. We first recognized semantic scene and then measured geometric structure, involving a delicate design of multistage refinement pipeline and a novel keypoint detection and description with global semantic awareness and local geometric verification, resulting in a practical visual geolocalization solution.

Gcg-net: Graph classification geolocation network
Yu Zhang, Shuhui Bu, Boni Hu, Pengcheng Han, Lean Weng, Shaocheng Xue
IEEE Transactions on Geoscience and Remote Sensing, 2023
IEEE paper

We designed a graph neural network (GNN) as the feature extraction network, and constructed a training framework based on image classification with a special data grouping strategy. Results demonstrate the effectiveness of our approach for UAV visual geolocation, especially for large scale applications.

Real-time dense point cloud generation and digital model construction of surface environment based on UAV platform
Boni Hu, Lin Chen, Bingli Xu, Shuhui Bu, Pengcheng Han, Li Kun, Zhenyu Xia, Ni Li, Ke Li, Xuefeng Cao, Gang Wan
National Remote Sensing Bulletin, 2023
paper

Our proposed geolocalization method is utilized for dense point cloud generation and digital model construction to survey washed-out and sediment areas, determining their location and extent accurately.

Change Detection for UAV Platforms

MDINet: Multi-Domain Incremental Network for Change Detection
Lean Weng, Wengqing Wang, Boni Hu, Pengcheng Han, Shaocheng Xue, Yu Zhang, Haowei Li, Jie Jin Shuhui Bu
IEEE Transactions on Geoscience and Remote Sensing, 2024
IEEE paper

We proposed a domain residual module, CD-DRU, decomposes the feature space into domain-specific and domain-shared parameters. This effectively mitigates catastrophic forgetting and outperforming existing Incremental Learning methods and state-of-the-art Change Detection techniques in remote sensing.

3D Reconstruction for Surgical Environment

3D Reconstruction of Deformable Colon Structures based on Preoperative Model and Deep Neural Network
Shuai Zhang, Liang Zhao, Shoudong Huang, Ruibin Ma, Boni Hu, Qi Hao,
IEEE International Conference on Robotics and Automation (ICRA 2021)
IEEE Paper

We utilized a preoperative colon model segmented from CT scans together with the colonoscopic images to achieve the 3D colon reconstruction, including dense depth estimation from monocular colonoscopic images using a deep neural network.

Deep learning assisted automatic intra-operative 3d aortic deformation reconstruction
Yanhao Zhang, Raphael Falque, Liang Zhao, Shoudong Huang, Boni Hu,
Medical Image Computing and Computer Assisted Intervention–MICCAI 2020
Springer paper

We provided a framework that reconstructed the live 3D aortic shape by fusing a 3D static pre-operative model and the 2D intra-operative fluoroscopic images.


Services
  • Reviewer of IEEE Transactions on Geoscience and Remote Sensing, ICPR, ICRA, ECCV, IROS, IEEE Transactions on Circuits and Systems for Video Technology.

This website is adapted from Jon Barron