TY - GEN
T1 - Fast Pose Estimation Method for Stacked Bolts
AU - Yuan, Bo
AU - Yin, Dong
AU - Li, Yuling
AU - Hou, Xudong
AU - Zhang, Mingfei
AU - Yu, Lezheng
AU - Zhao, Lijun
N1 - Publisher Copyright:
© 2024 IEEE.
PY - 2024
Y1 - 2024
N2 - This paper introduces a two-stage method for the pose estimation of bolts in complex stacked scenarios. The proposed approach is based on the open-source FPCC network and employs a custom dataset of stacked bolts, designed following the IPA stacked dataset generation methodology. To highlight the distinct features of bolts, we have adjusted the local coordinate system from the geometric center to the head of the hexagonal sections. During the instance generation phase of the fast clustering algorithm, we utilize DBSCAN and Mean-Shift techniques to refine the results of instance segmentation. Subsequently, SAC-IA and ICP are integrated for pose estimation, facilitating the alignment of the point cloud around the hexagonal bolt head. Our method enhances the speed of pose estimation while maintaining precision in stacked bolt scenarios.
AB - This paper introduces a two-stage method for the pose estimation of bolts in complex stacked scenarios. The proposed approach is based on the open-source FPCC network and employs a custom dataset of stacked bolts, designed following the IPA stacked dataset generation methodology. To highlight the distinct features of bolts, we have adjusted the local coordinate system from the geometric center to the head of the hexagonal sections. During the instance generation phase of the fast clustering algorithm, we utilize DBSCAN and Mean-Shift techniques to refine the results of instance segmentation. Subsequently, SAC-IA and ICP are integrated for pose estimation, facilitating the alignment of the point cloud around the hexagonal bolt head. Our method enhances the speed of pose estimation while maintaining precision in stacked bolt scenarios.
KW - Instance Segmentation
KW - Local Coordinate System
KW - Point Cloud Alignment
KW - Pose Estimation
KW - Stacked Bolts
UR - https://www.scopus.com/pages/publications/85200651935
U2 - 10.1109/CVIDL62147.2024.10604117
DO - 10.1109/CVIDL62147.2024.10604117
M3 - 会议稿件
AN - SCOPUS:85200651935
T3 - 2024 5th International Conference on Computer Vision, Image and Deep Learning, CVIDL 2024
SP - 577
EP - 581
BT - 2024 5th International Conference on Computer Vision, Image and Deep Learning, CVIDL 2024
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 5th International Conference on Computer Vision, Image and Deep Learning, CVIDL 2024
Y2 - 19 April 2024 through 21 April 2024
ER -