Generative regularization with latent topics for discriminative object recognition

• Verfasst von: Rubio, Jose C. [VerfasserIn]
• Verfasst von: Eigenstetter, Angela [VerfasserIn]
• Verfasst von: Ommer, Björn [VerfasserIn]
• Titel: Generative regularization with latent topics for discriminative object recognition
• Verf.angabe: Jose C. Rubio, Angela Eigenstetter, Björn Ommer
• E-Jahr: 2015
• Jahr: [2015]
• Umfang: 10 S.
• Illustrationen: Illustrationen
• Fussnoten: Available online 2 July 2015 ; Gesehen am 08.07.2020
• Titel Quelle: Enthalten in: Pattern recognition
• Ort Quelle: Amsterdam : Elsevier, 1968
• Jahr Quelle: 2015
• Band/Heft Quelle: 48(2015), 12, Seite 3871-3880
• DOI: doi:10.1016/j.patcog.2015.06.013
• Datenträger: Online-Ressource
• Sprache: eng
• Sach-SW: Latent SVM
• Sach-SW: Mixture models
• Sach-SW: Non-Negative Matrix Factorization
• Sach-SW: Part-based models
• Sach-SW: Visual object recognition
• K10plus-PPN: 1713927713

Abstract

Popular part-based approaches to recognition are currently limited to few localized parts, which only poorly represent the fine-scale details and large variability of object categories. Extending to hundreds of specific part detectors helps to capture peculiar characteristics but due to their specificity, for each object instance different parts will be helpful and others will yield noisy responses that actually impair classification. While training the part-based model, we thus need to learn which parts are relevant for which training instances. To automatically discover these latent topics of parts and instances we employ generative non-negative matrix factorization and seek topics with low reconstruction error. To assure recognition performance this generative approach is embedded within a discriminative latent max-margin procedure that separates classes while optimizing the latent topics. Consequently, generative reconstruction is regularizing discriminative classification, while the latter ensures that topics actually help in recognition. Experiments on PASCAL VOC demonstrate the recognition performance of our model as well as the construction of meaningful topics.