Deep unsupervised learning of visual similarities

• Verfasst von: Sanakoyeu, Artsiom [VerfasserIn]
• Verfasst von: Bautista, Miguel [VerfasserIn]
• Verfasst von: Ommer, Björn [VerfasserIn]
• Titel: Deep unsupervised learning of visual similarities
• Verf.angabe: Artsiom Sanakoyeu, Miguel A. Bautista, Björn Ommer
• E-Jahr: 2018
• Jahr: 31 January 2018
• Umfang: 13 S.
• Teil: volume:78
• Teil: year:2018
• Teil: pages:331-343
• Teil: extent:13
• Fussnoten: Gesehen am 16.05.2019
• Titel Quelle: Enthalten in: Pattern recognition
• Ort Quelle: Amsterdam : Elsevier, 1968
• Jahr Quelle: 2018
• Band/Heft Quelle: 78(2018), Seite 331-343
• DOI: doi:10.1016/j.patcog.2018.01.036
• Datenträger: Online-Ressource
• Sprache: eng
• Sach-SW: Deep learning
• Sach-SW: Human pose analysis
• Sach-SW: Object retrieval
• Sach-SW: Self-supervised learning
• Sach-SW: Visual similarity learning
• K10plus-PPN: 1665801212

Abstract

Exemplar learning of visual similarities in an unsupervised manner is a problem of paramount importance to computer vision. In this context, however, the recent breakthrough in deep learning could not yet unfold its full potential. With only a single positive sample, a great imbalance between one positive and many negatives, and unreliable relationships between most samples, training of Convolutional Neural networks is impaired. In this paper we use weak estimates of local similarities and propose a single optimization problem to extract batches of samples with mutually consistent relations. Conflicting relations are distributed over different batches and similar samples are grouped into compact groups. Learning visual similarities is then framed as a sequence of categorization tasks. The CNN then consolidates transitivity relations within and between groups and learns a single representation for all samples without the need for labels. The proposed unsupervised approach has shown competitive performance on detailed posture analysis and object classification.