Haonan Guo (Pole)

Recent Projects

AI+Remote Sensing suffers from poor performance due to the Lack of Prior Knowledge. l have developed several prior knowledge-guided methods for multiple tasks(e.g.building mapping, change detection).

Currently, I am working on the Remote Sensing ChatGPT project, developing an intelligent system for one-stop remote sensing image interpretation.

Remote Sensing ChatGPT

Paper(IGARSS) | Project Page | Blog | Zhihu

The potential of LLM in handling remote sensing interpretation has not yet been explored. Remote Sensing ChatGPT is an LLM-powered agent that utilizes ChatGPT to connect various AI models and solve complicated interpretation tasks. Users can simply send a remote sensing image with the corresponding request, and get the interpretation results as well as language feedback from Remote Sensing ChatGPT.

Metric-guided Change Detection

Paper 1| Paper 2(IEEE TIP)

I would like to say, DeepCL is super simple and effective! We complement the strong temporal modeling ability of metric learning with the prominent fitting ability of segmentation and propose a deep change feature learning (DeepCL) framework for robust and explainable CD. It is super simple in implementation and is effective!

Historical Building Footprint Update

Paper(RSE) | Project Page

Existing building contour databases need to be updated regularly to ensure their validity. We developed a novel algorithm to automatically update the existing building databases to their current status with minimal manual intervention. It is a promising approach for producing up-to-date building footprints in practical applications.(ESI HOT Paper,ESI Highly-cited Paper)

Building Boundary Refinement Mapping

Paper(ISPRS P&RS) | Project Page

Obstructions from nearby shadows or trees, varying shapes of rooftops, omission of small buildings, and varying scale of buildings hinder existing automated models for extracting sharp building boundaries.CBR-Net addressed these problems by introducing direction feild for boundary refinement. It achieved SOTA performance on several building footprint datasets.

Research

I'm interested in developing simple but effective models for remote sensing interpretation(e.g. segmentation and change detection) using geographical knowledge prior.

Published Papers:

1. Deep Building Footprint Update Network: A Semi-Supervised Method for Updating Existing Building Footprint from Bi-Temporal Remote Sensing Images
   Haonan Guo, Qian Shi, A. Marinoni, Bo Du, and Liangpei Zhang
   Remote Sensing of Environment |ESI Hot Paper, ESI Highly Cited Ppaer | paper

 

2. A Coarse-To-Fine Boundary Refinement Network for Building Footprint Extraction from Remote Sensing Imagery
   Haonan Guo, Bo Du, Liangpei Zhang, and Xin Su
   ISPRS Journal of Photogrammetry and Remote Sensing| ESI Highly Cited Paper | paper

 

3. Scene-Driven Multitask Parallel Attention Network for Building Extraction in High-Resolution Remote Sensing Images
    Haonan Guo, Qian Shi, Bo Du, Liangpei Zhang, Dongzhi Wang, and Huaxiang Ding,
   IEEE Transactions on Geoscience and Remote Sensing| ESI Highly Cited Paper | paper

 

4. SAAN: Similarity-aware attention flow network for change detection with VHR remote sensing images
   Haonan Guo, Xin Su, Chen Wu, Bo Du, and Liangpei Zhang
   IEEE Transactions on Image Processing | paper

 

5. On China's Earth Observation System: Mission, Vision and Application
   Deren Li, Mi Wang, Haonan Guo*, Wenjie Jin
   Geo-spatial Information Science | paper

 

6. Building-Road Collaborative Extraction from Remote Sensing Images via Cross-Task and Cross-Scale Interaction
   Haonan Guo, Xin Su, Chen Wu, Bo Du, and Liangpei Zhang
   IEEE Transactions on Geoscience and Remote Sensing | paper

 

7. Decoupling Semantic and Edge Representations for Building Footprint Extraction from Remote Sensing Images
   Haonan Guo, Xin Su, Chen Wu, Bo Du, and Liangpei Zhang
   IEEE Transactions on Geoscience and Remote Sensing | paper

 

8.  Scale-Robust Deep-Supervision Network for Mapping Building Footprints From High-Resolution Remote Sensing Images
   Haonan Guo, Xin Su, Shengkun Tang, Bo Du, and Liangpei Zhang
   IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing | paper

 

Review Papers

1.  Intelligence Driven Earth Observation Data Processing and Knowledge Discovery: Progress and Perspective
   Coming Soon~ | Nature Reviews Electrical Engineering   | by Prof. Liangpei Zhang

 

Submitted Papers:

1. Remote Sensing ChatGPT: Solving Remote Sensing Tasks with ChatGPT and Visual Models
   Haonan Guo, Xin Su, Chen Wu, Bo Du, Liangpei Zhang, Deren Li
   IGARSS 2024, submitted | paper

 

2. Expediting Building Footprint Segmentation from High-resolution Remote Sensing Images via progressive lenient supervision
   Haonan Guo, Bo Du, Chen Wu, Xin Su, Liangpei Zhang
   JAG, submitted | paper

 

3. DeepCL: Deep Change Feature Learning on Remote Sensing Images in the Metric Space
   Haonan Guo, Bo Du, Chen Wu, Xin Su, Chengxi Han, Liangpei Zhang
   IEEE TNNLS, submitted | paper

Awards

2023, Champion in Gaofen Challenge on Automated High-Resolution Earth Observation Image Interpretation

2023, Champion in the Building Footprint Extraction Competition in Aliyun Tianchi Competition

2023, Outstanding Student Leader by Wuhan University

2023, First Prize of the Wuhan University Excellent Academic Scholarship

2022, First Place Presentation in the International Graduate Workshop on GeoInformatics (IGWG)

2022, First prize in the National Graduate Academic Forum by National Remote Sensing Bulletin

2022, Second Prize of Academic Innovation Award by Wuhan University

2019, Baosteel Fellowship

2018, National Scholarship

2018, Second Prize in the National GIS Application Skills Competition