Xiaoxue Chen
Papers
6
Total Citations
72
H-Index
4
About
Xiaoxue Chen is a rising researcher in computer vision and robotics, whose work focuses on 3D scene understanding, task-oriented perception, and novel view synthesis. Her most influential contribution, “PQ-Transformer: Jointly Parsing 3D Objects and Layouts From Point Clouds” (37 citations), pioneers a unified transformer architecture that simultaneously detects objects and estimates room layouts from point clouds, overcoming the limitations of separate, task-specific networks. This work is foundational for robotic environment sensing and motion planning. Chen has also advanced semi-supervised and omni-supervised room layout estimation, addressing the critical challenge of data scarcity in 3D annotation. Her paper “TOIST: Task Oriented Instance Segmentation Transformer with Noun-Pronoun Distillation” (9 citations) introduces a novel approach to understanding verb-referenced actions, enabling robots to identify objects that afford specific tasks—a key step toward more intuitive human-robot interaction. Additionally, her work on NeRRF (4 citations) tackles the difficult problem of reconstructing transparent and specular objects by modeling refractive and reflective light paths, pushing the boundaries of neural radiance fields. With a growing citation record and a clear trajectory from 3D parsing to task-oriented vision, Chen is shaping the future of intelligent robotic perception.
Research Focus
Key Achievements
Top Papers
- 1PQ-Transformer: Jointly Parsing 3D Objects and Layouts From Point Clouds37 citations · 2022
- 2
- 3
- 4
- 5
- 6