Semantic matching by weakly supervised 2D point set registration

Tutkimustuotos: Artikkeli kirjassa/konferenssijulkaisussavertaisarvioitu

Tutkijat

Organisaatiot

Kuvaus

In this paper we address the problem of establishing correspondences between different instances of the same object. The problem is posed as finding the geometric transformation that aligns a given image pair. We use a convolutional neural network (CNN) to directly regress the parameters of the transformation model. The alignment problem is defined in the setting where an unordered set of semantic key-points per image are available, but, without the correspondence information. To this end we propose a novel loss function based on cyclic consistency that solves this 2D point set registration problem by inferring the optimal geometric transformation model parameters. We train and test our approach on a standard benchmark dataset Proposal-Flow (PF-PASCAL)[8]. The proposed approach achieves state-of-the-art results demonstrating the effectiveness of the method. In addition, we show our approach further benefits from additional training samples in PF-PASCAL generated by using category level information.

Yksityiskohdat

AlkuperäiskieliEnglanti
Otsikko2019 IEEE WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION (WACV)
TilaJulkaistu - 4 maaliskuuta 2019
OKM-julkaisutyyppiA4 Artikkeli konferenssijulkaisuussa
TapahtumaIEEE Winter Conference on Applications of Computer Vision - Waikoloa Village, Yhdysvallat
Kesto: 7 tammikuuta 201911 tammikuuta 2019
Konferenssinumero: 19

Conference

ConferenceIEEE Winter Conference on Applications of Computer Vision
LyhennettäWACV
MaaYhdysvallat
KaupunkiWaikoloa Village
Ajanjakso07/01/201911/01/2019

Lataa tilasto

Ei tietoja saatavilla

ID: 34750372