Himanshu Arora

Papers

Total Citations

H-Index

About

Himanshu Arora is a researcher advancing the frontiers of computer vision and robotics, with a primary focus on 3D shape understanding. His key research area addresses the critical problem of **multimodal shape completion**—the task of inferring complete 3D shapes from partial, often noisy inputs like scans, which is essential for applications in autonomous systems and augmented reality. Arora’s major contribution lies in pioneering the use of **Implicit Maximum Likelihood Estimation (IMLE)** for this challenge. Unlike traditional deterministic methods that produce a single, often blurry output, his work generates multiple plausible, high-fidelity completions, capturing the inherent ambiguity of real-world data. His 2022 paper, "Multimodal Shape Completion via Implicit Maximum Likelihood Estimation," has garnered 13 citations, establishing a new paradigm for handling occlusion and sparsity. An earlier 2021 paper on the same theme (6 citations) laid the groundwork for this approach. By enabling models to produce diverse and realistic shape hypotheses, Arora’s research directly improves the robustness of perception systems in cluttered or partially observed environments, marking him as an emerging leader in 3D generative modeling.