Qingxuan Lv

Papers

Total Citations

H-Index

About

Qingxuan Lv is a rising researcher in computer vision, with a focused specialization in underwater perception and semantic segmentation. Their work directly addresses the unique challenges of autonomous systems operating in degraded visual environments. Lv’s major contributions include the development of **WaterSAM**, a foundational model that adapts the Segment Anything Model (SAM) for robust underwater object segmentation, achieving 20 citations since its 2024 release. This work is critical for applications in robotic vision and augmented reality where standard models fail due to reduced visibility and low contrast. To further tackle the data scarcity problem in this domain, Lv introduced **UWStereo**, a large synthetic dataset for underwater stereo matching, which provides essential ground truth data for training deep learning models in complex aquatic settings. Earlier foundational work includes an **Embedded Attention Network** for semantic segmentation, which leverages self-attention mechanisms to capture long-range dependencies, enhancing accuracy for robot navigation. By bridging the gap between general-purpose vision models and the specific demands of underwater environments, Lv is paving the way for more reliable autonomous systems in marine exploration and robotics.