Files in this item

FilesDescriptionFormat

application/pdf

 Title: Recognition using visual phrases Author(s): Sadeghi, Mohammad Amin Advisor(s): Forsyth, David A. Department / Program: Computer Science Discipline: Computer Science Degree Granting Institution: University of Illinois at Urbana-Champaign Degree: M.S. Genre: Thesis Subject(s): Visual Phrase Phrasal Recognition Visual Composites Object Recognition Abstract: In this thesis I introduce visual phrases, complex visual composites like a person riding a horse''. Visual phrases often display significantly reduced visual complexity compared to their component objects, because the appearance of those objects can change profoundly when they participate in relations. I introduce a dataset suitable for phrasal recognition that uses familiar PASCAL object categories, and demonstrate significant experimental gains resulting from exploiting visual phrases. I show that a visual phrase detector significantly outperforms a baseline which detects component objects and reasons about relations, even though visual phrase training sets tend to be smaller than those for objects. I argue that any multi-class detection system must decode detector outputs to produce final results; this is usually done with non-maximum suppression. I describe a novel decoding procedure that can account accurately for local context without solving difficult inference problems. I show this decoding procedure outperforms the state of the art. Finally, I show that decoding a combination of phrasal and object detectors produces real improvements in detector results. Issue Date: 2012-06-27 URI: http://hdl.handle.net/2142/32060 Rights Information: Copyright 2012 Mohammad Amin Sadeghi and Ali Farhadi under Creative Commons Date Available in IDEALS: 2014-06-28 Date Deposited: 2012-05