Gui-Song Xia

About

Gui-Song Xia is a leading researcher at the intersection of computer vision, remote sensing, and autonomous robotics, with a particular focus on unmanned aerial vehicles (UAVs). His most significant contribution is the creation of the **UAVid dataset**, a large-scale benchmark for semantic segmentation of UAV imagery that has garnered over **367 citations** and become a foundational resource for the field, enabling advances in autonomous driving and robotic perception. Xia’s work extends to **event-based vision**, where he has pioneered methods for event-enhanced high-quality image recovery (112 citations) and robust keypoint detection using event-aligned features, addressing the limitations of conventional cameras under rapid motion and poor lighting. He has also advanced autonomous UAV manipulation, developing rotation-aware grasping systems that transform passive sensing into active 3D interaction. With a career spanning high-impact datasets, novel fusion techniques, and practical robotics applications, Xia’s research has shaped how machines perceive and interact with dynamic environments, making him a pivotal figure in modern computer vision and aerial robotics.

Research Focus

Key Achievements

Top Papers

1. 1
   UAVid: A semantic segmentation dataset for UAV imagery

   367 citations · 2020
2. 2
   Event Enhanced High-Quality Image Recovery

   112 citations · 2020
3. 3
   Information processing for unmanned aerial vehicles (UAVs) in surveying, mapping, and navigation

   16 citations · 2018
4. 4
   The UAVid Dataset for Video Semantic Segmentation

   10 citations · 2018
5. 5
   UAVid: A Semantic Segmentation Dataset for UAV Imagery

   9 citations · 2018
6. 6
   Toward Robust Keypoint Detection and Tracking: A Fusion Approach With Event-Aligned Image Features

   4 citations · 2024
7. 7
   Toward Autonomous Rotation-Aware Unmanned Aerial Grasping

   3 citations · 2018

Key Collaborators

Contact & Links