Khushi Gupta

Papers

Total Citations

H-Index

About

Khushi Gupta is a computer vision researcher whose work bridges deep learning and 3D spatial understanding, with a focus on indoor scene reconstruction. Her most cited paper, "DeepPerimeter: Indoor Boundary Estimation from Posed Monocular Sequences" (2019, 9 citations), introduces a novel deep learning pipeline that infers a complete indoor perimeter—essentially an exterior boundary map—from a sequence of posed RGB images. By leveraging robust deep methods for depth estimation and wall segmentation, Gupta's approach transforms ordinary monocular video into a 3D point cloud representing the room's structural outline, enabling applications in architecture, robotics, and augmented reality. This work stands out for its practical elegance: it sidesteps expensive LiDAR or multi-view setups, using only standard camera input to achieve accurate boundary mapping. While her citation count is modest, the paper's foundational contribution to indoor spatial reasoning has garnered attention in the computer vision community, particularly for its potential to simplify building information modeling (BIM) and autonomous navigation. Gupta's research exemplifies how deep learning can extract high-level geometric insights from everyday visual data, marking her as an emerging voice in scene understanding.