File Download
There are no files associated with this item.
Links for fulltext
(May Require Subscription)
- Publisher Website: 10.1007/978-3-030-58607-2_27
- Scopus: eid_2-s2.0-85097370768
Supplementary
-
Citations:
- Scopus: 0
- Appears in Collections:
Conference Paper: Mapping in a Cycle: Sinkhorn Regularized Unsupervised Learning for Point Cloud Shapes
Title | Mapping in a Cycle: Sinkhorn Regularized Unsupervised Learning for Point Cloud Shapes |
---|---|
Authors | |
Keywords | Point cloud Unsupervised learning Dense correspondence Cycle-consistency |
Issue Date | 2020 |
Publisher | Springer. |
Citation | Proceedings of the 16th European Conference on Computer Vision (ECCV), Online, Glasgow, UK, 23-28 August 2020, pt X, p. 455-472 How to Cite? |
Abstract | We propose an unsupervised learning framework with the pretext task of finding dense correspondences between point cloud shapes from the same category based on the cycle-consistency formulation. In order to learn discriminative pointwise features from point cloud data, we incorporate in the formulation a regularization term based on Sinkhorn normalization to enhance the learned pointwise mappings to be as bijective as possible. Besides, a random rigid transform of the source shape is introduced to form a triplet cycle to improve the model’s robustness against perturbations. Comprehensive experiments demonstrate that the learned pointwise features through our framework benefits various point cloud analysis tasks, e.g. partial shape registration and keypoint transfer. We also show that the learned pointwise features can be leveraged by supervised methods to improve the part segmentation performance with either the full training dataset or just a small portion of it. |
Persistent Identifier | http://hdl.handle.net/10722/293456 |
ISBN | |
Series/Report no. | Lecture Notes in Computer Science (LNCS), v. 12355 |
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Yang, L | - |
dc.contributor.author | Liu, W | - |
dc.contributor.author | Cui, Z | - |
dc.contributor.author | Chen, N | - |
dc.contributor.author | Wang, WP | - |
dc.date.accessioned | 2020-11-23T08:17:01Z | - |
dc.date.available | 2020-11-23T08:17:01Z | - |
dc.date.issued | 2020 | - |
dc.identifier.citation | Proceedings of the 16th European Conference on Computer Vision (ECCV), Online, Glasgow, UK, 23-28 August 2020, pt X, p. 455-472 | - |
dc.identifier.isbn | 9783030586065 | - |
dc.identifier.uri | http://hdl.handle.net/10722/293456 | - |
dc.description.abstract | We propose an unsupervised learning framework with the pretext task of finding dense correspondences between point cloud shapes from the same category based on the cycle-consistency formulation. In order to learn discriminative pointwise features from point cloud data, we incorporate in the formulation a regularization term based on Sinkhorn normalization to enhance the learned pointwise mappings to be as bijective as possible. Besides, a random rigid transform of the source shape is introduced to form a triplet cycle to improve the model’s robustness against perturbations. Comprehensive experiments demonstrate that the learned pointwise features through our framework benefits various point cloud analysis tasks, e.g. partial shape registration and keypoint transfer. We also show that the learned pointwise features can be leveraged by supervised methods to improve the part segmentation performance with either the full training dataset or just a small portion of it. | - |
dc.language | eng | - |
dc.publisher | Springer. | - |
dc.relation.ispartof | European Conference on Computer Vision (ECCV 2020) | - |
dc.relation.ispartofseries | Lecture Notes in Computer Science (LNCS), v. 12355 | - |
dc.subject | Point cloud | - |
dc.subject | Unsupervised learning | - |
dc.subject | Dense correspondence | - |
dc.subject | Cycle-consistency | - |
dc.title | Mapping in a Cycle: Sinkhorn Regularized Unsupervised Learning for Point Cloud Shapes | - |
dc.type | Conference_Paper | - |
dc.identifier.email | Yang, L: lyang125@hku.hk | - |
dc.identifier.email | Wang, WP: wenping@cs.hku.hk | - |
dc.identifier.authority | Wang, WP=rp00186 | - |
dc.description.nature | link_to_subscribed_fulltext | - |
dc.identifier.doi | 10.1007/978-3-030-58607-2_27 | - |
dc.identifier.scopus | eid_2-s2.0-85097370768 | - |
dc.identifier.hkuros | 318909 | - |
dc.identifier.volume | pt X | - |
dc.identifier.spage | 455 | - |
dc.identifier.epage | 472 | - |
dc.publisher.place | Cham | - |