Preprint
Aug 2026
A Geometry-Driven, Framework-Agnostic Optimization for Object Pose Estimation
Current object pose estimation research remains predominantly model-centric, focusing on architectural innovations and post-processing refinements. This paper introduces a data-centric optimization by proposing a novel, physically grounded rotation representation through principal axes alignment. Our method aligns the...
Wei Chen, Tao Zhen, Zhongchen Shi et al.
· 0 citations