Highly Accurate Boundary Detection and Grouping

Abstract : In this work we address boundary detection and boundary grouping. We first pursue a learning- based approach to boundary detection. For this (i) we leverage appearance and context information by extracting descriptors around edgels and use them as features for classification, (ii) we use discrimina- tive dimensionality reduction for efficiency and (iii) we use outlier-resilient boosting to deal with noise in the training set. We then introduce fractional-linear programming to optimize a grouping criterion that is expressed as a cost ratio. Our contributions are systematically evaluated on the Berkeley benchmark.
Type de document :
Communication dans un congrès
CVPR - IEEE Conf. on Computer Vision and Pattern Recognition, 2010, San Francisco, United States. pp.2520-2527, 2010
Liste complète des métadonnées

https://hal.inria.fr/hal-00857481
Contributeur : Iasonas Kokkinos <>
Soumis le : mardi 3 septembre 2013 - 15:46:06
Dernière modification le : vendredi 12 janvier 2018 - 11:24:39

Identifiants

  • HAL Id : hal-00857481, version 1

Collections

Citation

Iasonas Kokkinos. Highly Accurate Boundary Detection and Grouping. CVPR - IEEE Conf. on Computer Vision and Pattern Recognition, 2010, San Francisco, United States. pp.2520-2527, 2010. 〈hal-00857481〉

Partager

Métriques

Consultations de la notice

121