{"id":{"repo_id":"uiuc","oai_identifier":"oai:www.ideals.illinois.edu:2142/95297"},"canonical_url":"https://search.dev.ndltd.org/etd/uiuc/oai:www.ideals.illinois.edu:2142/95297","repository":{"repo_id":"uiuc","name":"University of Illinois - Urbana-Champaign","base_url":"https://www.ideals.illinois.edu/oai-pmh"},"display":{"title":"Learning to localize landmarks","abstract":"The world is full of tiny but useful objects such as the door handle of a car or the light switch in a room. Such objects are barely visible in an image and can be well approximated by a single point. We refer to these small objects as landmarks in addition to the more common usage of the term to refer to the anatomical or facial landmarks. Landmark localization refers to the detection of one or more such landmarks in an image. In this dissertation, we describe methods for localizing such landmarks in images. Automatically localizing these landmarks in images is hard as they usually don’t have a distinctive appearance of their own. They are largely defined by their context. Absence of local appearance necessitates effective modeling of context to achieve good localization performance. This context can be explicit in the form of other landmarks that are spatially related or implicit in the form of certain recurring and informative patterns that may not have a semantic label. We describe methods that model both explicit and implicit context to improve landmark localization performance. Localization performance is tied to the underlying learning machinery being used. Deep neural networks have proven to be quite successful for computer vision applications and this dissertation employs them as the underlying learning machine. We describe a method that uses a deep neural network with residual architecture, a recently proposed architecture for classification, and improves its performance with a novel stochastic training method called Swapout.","abstract_html":"The world is full of tiny but useful objects such as the door handle of a car or the light switch in a room. Such objects are barely visible in an image and can be well approximated by a single point. We refer to these small objects as landmarks in addition to the more common usage of the term to refer to the anatomical or facial landmarks. Landmark localization refers to the detection of one or more such landmarks in an image. In this dissertation, we describe methods for localizing such landmarks in images. Automatically localizing these landmarks in images is hard as they usually don’t have a distinctive appearance of their own. They are largely defined by their context. Absence of local appearance necessitates effective modeling of context to achieve good localization performance. This context can be explicit in the form of other landmarks that are spatially related or implicit in the form of certain recurring and informative patterns that may not have a semantic label. We describe methods that model both explicit and implicit context to improve landmark localization performance. Localization performance is tied to the underlying learning machinery being used. Deep neural networks have proven to be quite successful for computer vision applications and this dissertation employs them as the underlying learning machine. We describe a method that uses a deep neural network with residual architecture, a recently proposed architecture for classification, and improves its performance with a novel stochastic training method called Swapout.","abstract_has_math":false,"creators":["Singh, Saurabh"],"institution":"University of Illinois at Urbana-Champaign","degree_name":"Ph.D.","degree_level":"Dissertation","degree_discipline":"Computer Science","degree_department":null,"school":null,"contributors":["Forsyth, David","Hoiem, Derek","Lazebnik, Svetlana","Ramanan, Deva K","Girshick, Ross B"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2017,"date_issued":"2017-03-01T15:46:14Z","date_published":"2017-03-01T15:46:14Z","updated_at":"2026-07-22T22:26:37Z","subjects":["Keypoint Localization","Pose estimation","Swapout"],"languages":["en"],"rights":["Copyright 2016 Saurabh Singh"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/2142/95297","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Forsyth, David","Hoiem, Derek","Lazebnik, Svetlana","Ramanan, Deva K","Girshick, Ross B"]},{"key":"dc:creator","label":"Author","values":["Singh, Saurabh"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2017-03-01T15:46:14Z","2016-10-19","2016-12"]},{"key":"dc:type","label":"Dc Type","values":["text"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Computer Science"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Dissertation"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Ph.D."]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Illinois at Urbana-Champaign"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Keypoint Localization","Pose estimation","Swapout"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright 2016 Saurabh Singh"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["http://hdl.handle.net/2142/95297"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["The world is full of tiny but useful objects such as the door handle of a car or the light switch in a room. Such objects are barely visible in an image and can be well approximated by a single point. We refer to these small objects as landmarks in addition to the more common usage of the term to refer to the anatomical or facial landmarks. Landmark localization refers to the detection of one or more such landmarks in an image. In this dissertation, we describe methods for localizing such landmarks in images. Automatically localizing these landmarks in images is hard as they usually don’t have a distinctive appearance of their own. They are largely defined by their context. Absence of local appearance necessitates effective modeling of context to achieve good localization performance. This context can be explicit in the form of other landmarks that are spatially related or implicit in the form of certain recurring and informative patterns that may not have a semantic label. We describe methods that model both explicit and implicit context to improve landmark localization performance. Localization performance is tied to the underlying learning machinery being used. Deep neural networks have proven to be quite successful for computer vision applications and this dissertation employs them as the underlying learning machine. We describe a method that uses a deep neural network with residual architecture, a recently proposed architecture for classification, and improves its performance with a novel stochastic training method called Swapout.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2017-02-28 without embargo terms","The student, Saurabh Singh, accepted the attached license on 2016-10-17 at 14:45.","The student, Saurabh Singh, submitted this Dissertation for approval on 2016-10-17 at 14:57.","This Dissertation was approved for publication on 2016-10-19 at 11:44.","DSpace SAF Submission Ingestion Package generated from Vireo submission #10191 on 2017-02-28 at 14:46:36","Made available in DSpace on 2017-03-01T15:46:14Z (GMT). No. of bitstreams: 2 SINGH-DISSERTATION-2016.pdf: 8018484 bytes, checksum: fbdd26f425d9bea881fa0cac013e984d (MD5) LICENSE.txt: 4210 bytes, checksum: 2e35cc89c6af434a4f6bad2c5abf3baa (MD5) Previous issue date: 2016-10-19"]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Learning to localize landmarks"]}]}],"canonical_facts":{"dc:contributor":["Forsyth, David","Hoiem, Derek","Lazebnik, Svetlana","Ramanan, Deva K","Girshick, Ross B"],"dc:creator":["Singh, Saurabh"],"dc:date":["2017-03-01T15:46:14Z","2016-10-19","2016-12"],"dc:description":["The world is full of tiny but useful objects such as the door handle of a car or the light switch in a room. Such objects are barely visible in an image and can be well approximated by a single point. We refer to these small objects as landmarks in addition to the more common usage of the term to refer to the anatomical or facial landmarks. Landmark localization refers to the detection of one or more such landmarks in an image. In this dissertation, we describe methods for localizing such landmarks in images. Automatically localizing these landmarks in images is hard as they usually don’t have a distinctive appearance of their own. They are largely defined by their context. Absence of local appearance necessitates effective modeling of context to achieve good localization performance. This context can be explicit in the form of other landmarks that are spatially related or implicit in the form of certain recurring and informative patterns that may not have a semantic label. We describe methods that model both explicit and implicit context to improve landmark localization performance. Localization performance is tied to the underlying learning machinery being used. Deep neural networks have proven to be quite successful for computer vision applications and this dissertation employs them as the underlying learning machine. We describe a method that uses a deep neural network with residual architecture, a recently proposed architecture for classification, and improves its performance with a novel stochastic training method called Swapout.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2017-02-28 without embargo terms","The student, Saurabh Singh, accepted the attached license on 2016-10-17 at 14:45.","The student, Saurabh Singh, submitted this Dissertation for approval on 2016-10-17 at 14:57.","This Dissertation was approved for publication on 2016-10-19 at 11:44.","DSpace SAF Submission Ingestion Package generated from Vireo submission #10191 on 2017-02-28 at 14:46:36","Made available in DSpace on 2017-03-01T15:46:14Z (GMT). No. of bitstreams: 2 SINGH-DISSERTATION-2016.pdf: 8018484 bytes, checksum: fbdd26f425d9bea881fa0cac013e984d (MD5) LICENSE.txt: 4210 bytes, checksum: 2e35cc89c6af434a4f6bad2c5abf3baa (MD5) Previous issue date: 2016-10-19"],"dc:format":["application/pdf"],"dc:identifier":["http://hdl.handle.net/2142/95297"],"dc:language":["en"],"dc:rights":["Copyright 2016 Saurabh Singh"],"dc:subject":["Keypoint Localization","Pose estimation","Swapout"],"dc:title":["Learning to localize landmarks"],"dc:type":["text"],"thesis:degree_discipline":["Computer Science"],"thesis:degree_level":["Dissertation"],"thesis:degree_name":["Ph.D."],"thesis:institution_name":["University of Illinois at Urbana-Champaign"]},"updated_at":"2026-07-22T22:26:37Z"}