Vage Taamazyan

Papers

Total Citations

137

H-Index

About

Vage Taamazyan is a computer vision researcher whose work sits at the intersection of physics-based vision, polarization imaging, and industrial robotics. His most influential contribution, the 2020 paper *"Deep Polarization Cues for Transparent Object Segmentation"* (130 citations), tackles one of the field’s most stubborn open problems: segmenting transparent objects that lack their own texture. By reframing this challenge through the lens of light polarization—specifically the rotation of light upon transmission—Taamazyan introduced a novel physical cue that deep learning models can exploit, enabling robust segmentation where traditional RGB-based methods fail. This work has become a foundational reference for researchers working on transparent and reflective object perception. More recently, Taamazyan has turned his attention to industrial-grade 6DoF pose estimation, co-authoring *"Towards Co-Evaluation of Cameras, HDR, and Algorithms for Industrial-Grade 6DoF Pose Estimation"* (2024), which introduces the Industrial Plenoptic Dataset (IPD)—the first benchmark designed for the simultaneous evaluation of camera hardware, HDR imaging, and pose estimation algorithms. This dataset addresses the critical gap between academic benchmarks and the reliability standards required for mass deployment in industrial robotics, positioning Taamazyan as a key figure bridging fundamental vision science with real-world automation.