Yasunori Hirakawa

Papers

Total Citations

H-Index

About

Yasunori Hirakawa is a researcher focused on advancing computer vision and robotics through innovative approaches to camera pose estimation. His work bridges the gap between traditional handcrafted feature methods and modern deep learning techniques, addressing critical challenges in how machines perceive and navigate their environments. In his 2022 paper, "Camera Attitude Estimation by Neural Network Using Classification Network Method Instead of Numerical Regression" (4 citations), Hirakawa introduced a novel framework that treats pose estimation as a classification problem—inspired by optical character recognition—enabling highly maneuverable terrestrial robots to accurately estimate their posture while freely tilting their upper bodies. Complementing this, his work "Handcrafted Features Can Help End-to-End Pose Estimation Using CNN" (2 citations) tackles the persistent issue of uncertainty in low-texture environments, demonstrating that integrating traditional handcrafted features with convolutional neural networks significantly improves accuracy. By systematically addressing the limitations of purely end-to-end estimators, Hirakawa’s research offers practical solutions for robust robot localization and navigation. His contributions are particularly valuable for students and engineers developing autonomous systems that must operate reliably in visually challenging conditions.