Please use this identifier to cite or link to this item: http://hdl.handle.net/2440/116856
Citations
Scopus Web of Science® Altmetric
?
?
Type: Conference paper
Title: Efficient dense point cloud object reconstruction using deformation vector fields
Author: Li, K.
Pham, T.
Zhan, H.
Reid, I.
Citation: Computer Vision - ECCV 2018 : 15th European Conference. Proceedings, Part XII, 2018 / Ferrari, V. (ed./s), vol.11216 LNCS, pp.508-524
Issue Date: 2018
Series/Report no.: Lecture Notes in Computer Science; 11216
ISBN: 9783030012571
ISSN: 0302-9743
1611-3349
Conference Name: European Conference on Computer Vision (ECCV) (08 Sep 2018 - 14 Sep 2018 : Munich, Germany)
Statement of
Responsibility: 
Kejie Li, Trung Pham, Huangying Zhan, Ian Reid
Abstract: Some existing CNN-based methods for single-view 3D object reconstruction represent a 3D object as either a 3D voxel occupancy grid or multiple depth-mask image pairs. However, these representations are inefficient since empty voxels or background pixels are wasteful. We propose a novel approach that addresses this limitation by replacing masks with “deformation-fields”. Given a single image at an arbitrary viewpoint, a CNN predicts multiple surfaces, each in a canonical location relative to the object. Each surface comprises a depth-map and corresponding deformation-field that ensures every pixel-depth pair in the depth-map lies on the object surface. These surfaces are then fused to form the full 3D shape. During training we use a combination of per-view loss and multi-view losses. The novel multi-view loss encourages the 3D points back-projected from a particular view to be consistent across views. Extensive experiments demonstrate the efficiency and efficacy of our method on single-view 3D object reconstruction.
Keywords: 3D object reconstruction; dense point clouds; deep learning
Rights: © Springer Nature Switzerland AG 2018
RMID: 0030102400
DOI: 10.1007/978-3-030-01258-8_31
Grant ID: http://purl.org/au-research/grants/arc/FL130100102
http://purl.org/au-research/grants/arc/CE140100016
Appears in Collections:Australian Institute for Machine Learning publications
Computer Science publications

Files in This Item:
There are no files associated with this item.


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.