Relative Attributes 专业英语论文.docxVIP

Relative Attributes AbstractHuman-nameable visual “attributes” can benefit various recognition tasks. However, existing techniques restrict these properties to categorical labels (for example, a person is ‘smiling’ or not, a scene is ‘dry’ or not), and thus fail to capture more general semantic relationships. We propose to model relative attributes. Given training data stating how object/scene categories relate according to different attributes, we learn a ranking function per attribute. The learned ranking functions predict the relative strength of each property in novel images. We then build a generative model over the joint space of attribute ranking outputs, and propose a novel form of zero-shot learning in which the supervisor relates the unseen object category to previously seen objects via attributes (for example, ‘bears are furrier than giraffes’). We further show how the proposed relative attributes enable richer textual descriptions for new images, which in practice are more precise for human interpretation. We demonstrate the approach on datasets of faces and natural scenes, and show its clear advantages over traditional binary attribute prediction for these new tasks.Keywords-relative; attributes; zero-shot learningIntroductionWhile traditional visual recognition approaches map low-level image features directly to object category labels, recent work proposes models using visual attributes [1-8]. Attributes are properties observable in images that have human-designated names (e.g., ‘striped’, ‘four-legged’),and they are valuable as a new semantic cue in various problems. For example, researchers have shown their impact for strengthening facial verification [5], object recognition, generating descriptions of unfamiliar objects [1], and to facilitate “zero-shot” transfer learning [2], where one trains a classifier for an unseen object simply by specifying which attributes it has.A.Problem: Most existing work focuses wholly on attributes as binary predicates