OmniPose6D: Towards Short-Term Object Pose Tracking in Dynamic Scenes from Monocular RGB

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Lin, Yunzhi, Zhao, Yipu, Chu, Fu-Jen, Chen, Xingyu, Wang, Weiyao, Tang, Hao, Vela, Patricio A., Feiszli, Matt, Liang, Kevin
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866918110718590976
author Lin, Yunzhi
Zhao, Yipu
Chu, Fu-Jen
Chen, Xingyu
Wang, Weiyao
Tang, Hao
Vela, Patricio A.
Feiszli, Matt
Liang, Kevin
author_facet Lin, Yunzhi
Zhao, Yipu
Chu, Fu-Jen
Chen, Xingyu
Wang, Weiyao
Tang, Hao
Vela, Patricio A.
Feiszli, Matt
Liang, Kevin
contents To address the challenge of short-term object pose tracking in dynamic environments with monocular RGB input, we introduce a large-scale synthetic dataset OmniPose6D, crafted to mirror the diversity of real-world conditions. We additionally present a benchmarking framework for a comprehensive comparison of pose tracking algorithms. We propose a pipeline featuring an uncertainty-aware keypoint refinement network, employing probabilistic modeling to refine pose estimation. Comparative evaluations demonstrate that our approach achieves performance superior to existing baselines on real datasets, underscoring the effectiveness of our synthetic dataset and refinement technique in enhancing tracking precision in dynamic contexts. Our contributions set a new precedent for the development and assessment of object pose tracking methodologies in complex scenes.
format Preprint
id arxiv_https___arxiv_org_abs_2410_06694
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle OmniPose6D: Towards Short-Term Object Pose Tracking in Dynamic Scenes from Monocular RGB
Lin, Yunzhi
Zhao, Yipu
Chu, Fu-Jen
Chen, Xingyu
Wang, Weiyao
Tang, Hao
Vela, Patricio A.
Feiszli, Matt
Liang, Kevin
Computer Vision and Pattern Recognition
Robotics
To address the challenge of short-term object pose tracking in dynamic environments with monocular RGB input, we introduce a large-scale synthetic dataset OmniPose6D, crafted to mirror the diversity of real-world conditions. We additionally present a benchmarking framework for a comprehensive comparison of pose tracking algorithms. We propose a pipeline featuring an uncertainty-aware keypoint refinement network, employing probabilistic modeling to refine pose estimation. Comparative evaluations demonstrate that our approach achieves performance superior to existing baselines on real datasets, underscoring the effectiveness of our synthetic dataset and refinement technique in enhancing tracking precision in dynamic contexts. Our contributions set a new precedent for the development and assessment of object pose tracking methodologies in complex scenes.
title OmniPose6D: Towards Short-Term Object Pose Tracking in Dynamic Scenes from Monocular RGB
topic Computer Vision and Pattern Recognition
Robotics
url https://arxiv.org/abs/2410.06694