Hoang Thanh Le

Papers

Total Citations

H-Index

About

Hoang Thanh Le is a rising researcher in computer vision and 3D geometric deep learning, whose work addresses fundamental challenges in reconstructing complete and accurate 3D representations from partial sensor data. Le’s primary research focuses on point cloud completion, a critical task for enabling robust perception in robotics, autonomous navigation, and augmented reality. His most notable contribution, the CenFormer architecture, introduces a novel transformer-based network that generates precise centroid representations to guide the completion of sparse and occluded point clouds. This work, published in 2025, has already garnered early citations, signaling its potential impact on the field. By tackling the inherent limitations of 3D scanners—such as occlusions, limited viewpoints, and sensor noise—Le’s research directly improves the reliability of downstream applications that depend on high-fidelity 3D models. His approach represents a significant step forward in leveraging attention mechanisms for geometric reasoning, offering a more principled alternative to traditional completion methods. As his work continues to gain traction, Hoang Thanh Le is establishing himself as a promising voice in the next generation of 3D vision researchers.