Charu Sharma

Indian Institute of Technology Hyderabad, International Institute of Information Technology, Hyderabad

Papers

Total Citations

H-Index

About

Charu Sharma is a researcher working at the intersection of 3D computer vision, machine learning, and robotics, with a particular focus on point cloud understanding and domain adaptation. His work addresses some of the most pressing challenges in deploying deep learning models in real-world 3D environments, where labeled data is scarce and domain shifts are common. Sharma's most influential contribution, "Self-Supervised Few-Shot Learning on Point Clouds" (2020, 47 citations), pioneered approaches for learning robust 3D representations with minimal labeled data — a critical advancement for applications in autonomous driving, robotics, and shape synthesis. Building on this foundation, his work on contrastive learning combined with optimal transport for unsupervised domain adaptation (2024, 11 citations) offers practical solutions to bridging the gap between synthetic and real-world point cloud data. More recently, Sharma has pushed toward the frontier of embodied AI, exploring how Large Language Models can be integrated with 3D vision to enable more intelligent robotic perception and autonomy. His applied work also extends to infrastructure inspection, developing hyper-dense CNNs for wastewater pipe abnormality detection. Across these diverse contributions, Sharma demonstrates a consistent commitment to making 3D perception more robust, generalizable, and practically deployable.