Prateek Singhal

Georgia Institute of Technology, Indian Institute of Technology Hyderabad

Papers

Total Citations

H-Index

About

Prateek Singhal is a computer vision researcher whose work sits at the intersection of 3D scene reconstruction, semantic understanding, and motion analysis. His research primarily focuses on enabling machines to interpret complex visual environments — particularly dynamic urban scenes — with greater accuracy and contextual awareness. Singhal's most influential contribution, "Dynamic body VSLAM with semantic constraints" (2015, 40 citations), tackles one of the field's persistent challenges: reconstructing urban environments in the presence of moving objects. By integrating semantic constraints into Visual Simultaneous Localization and Mapping (VSLAM), his work significantly advances the reliability of large-scale 3D reconstruction pipelines that traditionally assume static scenes. Complementing this, his work on "Semantic Motion Segmentation Using Dense CRF Formulation" (2014, 22 citations) demonstrates a sophisticated bidirectional relationship between motion cues and semantic labeling — showing that each can meaningfully enhance the other, a novel insight that bridges two previously siloed research streams. His additional work on multi-plane detection from image pairs reflects a broader interest in structured geometric scene understanding. Collectively, Singhal's research contributes foundational tools for autonomous navigation, robotics, and intelligent scene parsing in real-world, unconstrained environments.