Skip to Main content Skip to Navigation
Conference papers

The Canonically Posed 3D Objects Dataset

Panagiotis Papadakis 1
1 Lagadic - Visual servoing in robotics, computer vision, and augmented reality
CRISAM - Inria Sophia Antipolis - Méditerranée , Inria Rennes – Bretagne Atlantique , IRISA-D5 - SIGNAUX ET IMAGES NUMÉRIQUES, ROBOTIQUE
Abstract : Shape matching methodologies of generic 3D objects are conventionally preceded by a pose normalization stage, that transforms objects to a canonical coordinate frame wherein feature extraction and shape matching is performed. Arguably, the canonical pose of a 3D object depends not only on its geometry but also on its semantic meaning, a characteristic that generally complicates the extraction of ground truth data. This paper introduces the first ground-truth dataset of 3D objects that allows an objective evaluation of methods which obtain the canonical pose of objects within extrinsic space. By virtue of the protocol that was followed to assemble the dataset, 3D objects of the same class share a fixed pose in terms of object center, scale and rotation while undergoing diverse shape deformations. The dataset is publicly disclosed and relevant use cases are discussed.
Document type :
Conference papers
Complete list of metadata

Cited literature [6 references]  Display  Hide  Download

https://hal.inria.fr/hal-00975099
Contributor : Panagiotis Papadakis <>
Submitted on : Tuesday, April 8, 2014 - 10:13:39 AM
Last modification on : Friday, July 10, 2020 - 4:01:17 PM
Long-term archiving on: : Tuesday, July 8, 2014 - 11:00:44 AM

File

CAPOD-3DOR2014.pdf
Publisher files allowed on an open archive

Identifiers

Citation

Panagiotis Papadakis. The Canonically Posed 3D Objects Dataset. Eurographics Workshop on 3D Object Retrieval, Apr 2014, Strasbourg, France. pp.33-36, ⟨10.2312/3dor.20141047⟩. ⟨hal-00975099⟩

Share

Metrics

Record views

523

Files downloads

471