Siddharth Mahendran

Papers

Total Citations

H-Index

About

Siddharth Mahendran is a computer vision researcher whose work centers on the fundamental challenge of 3D pose estimation from 2D images—a critical capability for autonomous navigation, robot manipulation, and augmented reality. His most influential contribution, the 2017 paper "3D Pose Regression Using Convolutional Neural Networks" (67 citations), challenged the prevailing paradigm of treating pose estimation as a classification problem. Mahendran demonstrated that direct regression of continuous 3D pose parameters using CNNs is not only feasible but often more accurate than discretized classification approaches. He further advanced this framework in his 2018 follow-up work, "A Mixed Classification-Regression Framework for 3D Pose Estimation from 2D Images" (24 citations), which elegantly combined the strengths of both paradigms to handle the inherent ambiguity in single-image pose recovery. By pioneering end-to-end regression methods for 3D pose, Mahendran has helped shift the field toward more elegant and geometrically meaningful solutions. His work provides a foundation for modern approaches that treat pose as a continuous quantity, enabling more precise and robust 3D understanding for real-world systems.