What question did this study set out to answer?

Develop an RGB-D cascaded fusion framework to accurately estimate the 3D geometric properties of irregular coal and gangue fragments without relying on predefined shape models.

September 17, 2026SeparationsOpen Access

RGB-D Semantic-Guided 3D Geometric Estimation for Intelligent Coal Gangue Separation

View Full Paper

Ask AI

Bookmark

Authors

ZSZhiyuan Sun ZZZhongjun Zhou TSTianyang Sun

Discussion

Member takes

Overview

Computational study demonstrates accurate 3D geometric estimation for irregular coal and gangue fragments, indicating reliable perception for automated robotic sorting.

Key Points

Develop an RGB-D cascaded fusion framework to accurately estimate the 3D geometric properties of irregular coal and gangue fragments without relying on predefined shape models.
Designed a detection-segmentation-reconstruction pipeline utilizing 2D object detection for semantic priors and zero-shot segmentation to produce pixel-level masks.
Used segmented masks to isolate depth information, reconstruct local partial point clouds, and estimate centroid position, oriented bounding box dimensions, visible volume, and principal orientation on a specialized RGB-D dataset.
Achieved a centroid localization error of 14.22 mm, a bounding box dimension error of 9.34%, and a visible volume error of 11.3%.
Demonstrated marked improvements in geometric estimation accuracy when using semantic-guided depth reconstruction over conventional detection-box-based methods.

Cite This Study

Sun et al. (2026) studied this question.

synapsesocial.com/papers/6aabb7975f706d05830e6caa https://doi.org/10.3390/separations13090263

View Full Paper

Ask AI

Bookmark

September 17, 2026SeparationsOpen Access

RGB-D Semantic-Guided 3D Geometric Estimation for Intelligent Coal Gangue Separation

View Full Paper

Ask AI

Bookmark

Authors

ZSZhiyuan Sun ZZZhongjun Zhou TSTianyang Sun

Discussion

Member takes

Overview

Computational study demonstrates accurate 3D geometric estimation for irregular coal and gangue fragments, indicating reliable perception for automated robotic sorting.

Key Points

Develop an RGB-D cascaded fusion framework to accurately estimate the 3D geometric properties of irregular coal and gangue fragments without relying on predefined shape models.
Designed a detection-segmentation-reconstruction pipeline utilizing 2D object detection for semantic priors and zero-shot segmentation to produce pixel-level masks.
Used segmented masks to isolate depth information, reconstruct local partial point clouds, and estimate centroid position, oriented bounding box dimensions, visible volume, and principal orientation on a specialized RGB-D dataset.
Achieved a centroid localization error of 14.22 mm, a bounding box dimension error of 9.34%, and a visible volume error of 11.3%.
Demonstrated marked improvements in geometric estimation accuracy when using semantic-guided depth reconstruction over conventional detection-box-based methods.

Cite This Study

Sun et al. (2026) studied this question.

synapsesocial.com/papers/6aabb7975f706d05830e6caa https://doi.org/10.3390/separations13090263

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark