{"id":{"repo_id":"uiuc","oai_identifier":"oai:www.ideals.illinois.edu:2142/106175"},"canonical_url":"https://search.dev.ndltd.org/etd/uiuc/oai:www.ideals.illinois.edu:2142/106175","repository":{"repo_id":"uiuc","name":"University of Illinois - Urbana-Champaign","base_url":"https://www.ideals.illinois.edu/oai-pmh"},"display":{"title":"Reducing false positives for object detection","abstract":"In this thesis, we analyze failure cases of state-of-the-art detectors and observe that most hard false positives result from classification instead of localization and they have a large negative impact on the performance of object detectors. We conjecture three factors that lie behind hard false positives, and we confirm the conjecture with experiments that prove the following: (1) Shared feature representation is not optimal due to the mismatched goals of feature learning for classification and localization; (2) large receptive field for different scales leads to redundant context information for small objects; (3) multi-task learning helps, yet optimization of the multi-task loss may prove sub-optimal for individual tasks. We demonstrate the potential power of detector classification by a simple, effective, and widely applicable Decoupled Classification Refinement (DCR) network. In particular, DCR places a separate classification network in parallel with the localization network (base detector). With ROI pooling placed on the early stage of the classification network, we enforce an adaptive receptive field in DCR. During training, DCR samples hard false positives from the base detector and trains a strong classifier to refine classification results. During testing, DCR refines all boxes from the base detector. Experiments show competitive results on PASCAL VOC and COCO without any bells and whistles. Our codes are available at: https://github.com/bowenc0221/Decoupled-Classification-Refinement.","abstract_html":"In this thesis, we analyze failure cases of state-of-the-art detectors and observe that most hard false positives result from classification instead of localization and they have a large negative impact on the performance of object detectors. We conjecture three factors that lie behind hard false positives, and we confirm the conjecture with experiments that prove the following: (1) Shared feature representation is not optimal due to the mismatched goals of feature learning for classification and localization; (2) large receptive field for different scales leads to redundant context information for small objects; (3) multi-task learning helps, yet optimization of the multi-task loss may prove sub-optimal for individual tasks. We demonstrate the potential power of detector classification by a simple, effective, and widely applicable Decoupled Classification Refinement (DCR) network. In particular, DCR places a separate classification network in parallel with the localization network (base detector). With ROI pooling placed on the early stage of the classification network, we enforce an adaptive receptive field in DCR. During training, DCR samples hard false positives from the base detector and trains a strong classifier to refine classification results. During testing, DCR refines all boxes from the base detector. Experiments show competitive results on PASCAL VOC and COCO without any bells and whistles. Our codes are available at: https://github.com/bowenc0221/Decoupled-Classification-Refinement.","abstract_has_math":false,"creators":["Cheng, Bowen"],"institution":"University of Illinois at Urbana-Champaign","degree_name":"M.S.","degree_level":"Thesis","degree_discipline":"Electrical & Computer Engr","degree_department":null,"school":null,"contributors":["Huang, Thomas S."],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2020,"date_issued":"2020-03-02T21:58:05Z","date_published":"2020-03-02T21:58:05Z","updated_at":"2026-07-22T22:24:45Z","subjects":["Object Detection","Deep Learning"],"languages":["en"],"rights":["Copyright 2019 Bowen Cheng"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/2142/106175","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Huang, Thomas S."]},{"key":"dc:creator","label":"Author","values":["Cheng, Bowen"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2020-03-02T21:58:05Z","2019-10-16","2019-12"]},{"key":"dc:type","label":"Dc Type","values":["text"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Electrical & Computer Engr"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["M.S."]},{"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":["Object Detection","Deep Learning"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright 2019 Bowen Cheng"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["http://hdl.handle.net/2142/106175"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["In this thesis, we analyze failure cases of state-of-the-art detectors and observe that most hard false positives result from classification instead of localization and they have a large negative impact on the performance of object detectors. We conjecture three factors that lie behind hard false positives, and we confirm the conjecture with experiments that prove the following: (1) Shared feature representation is not optimal due to the mismatched goals of feature learning for classification and localization; (2) large receptive field for different scales leads to redundant context information for small objects; (3) multi-task learning helps, yet optimization of the multi-task loss may prove sub-optimal for individual tasks. We demonstrate the potential power of detector classification by a simple, effective, and widely applicable Decoupled Classification Refinement (DCR) network. In particular, DCR places a separate classification network in parallel with the localization network (base detector). With ROI pooling placed on the early stage of the classification network, we enforce an adaptive receptive field in DCR. During training, DCR samples hard false positives from the base detector and trains a strong classifier to refine classification results. During testing, DCR refines all boxes from the base detector. Experiments show competitive results on PASCAL VOC and COCO without any bells and whistles. Our codes are available at: https://github.com/bowenc0221/Decoupled-Classification-Refinement.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2020-02-28 without embargo terms","The student, Bowen Cheng, accepted the attached license on 2019-10-16 at 13:52.","The student, Bowen Cheng, submitted this Thesis for approval on 2019-10-16 at 13:59.","This Thesis was approved for publication on 2019-10-16 at 16:43.","DSpace SAF Submission Ingestion Package generated from Vireo submission #14500 on 2020-02-28 at 17:13:00","Made available in DSpace on 2020-03-02T21:58:05Z (GMT). No. of bitstreams: 3 CHENG-THESIS-2019.pdf: 3339997 bytes, checksum: 6aed1a4f984eb5c6e93565da85e10b3e (MD5) LICENSE.txt: 4208 bytes, checksum: 49dcaf80b3c39981f08301ac89ec45d7 (MD5) RightsLink Printable License.pdf: 115296 bytes, checksum: 08215077c5fa568e4fb59fa8ceec94e1 (MD5) Previous issue date: 2019-10-16"]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Reducing false positives for object detection"]}]}],"canonical_facts":{"dc:contributor":["Huang, Thomas S."],"dc:creator":["Cheng, Bowen"],"dc:date":["2020-03-02T21:58:05Z","2019-10-16","2019-12"],"dc:description":["In this thesis, we analyze failure cases of state-of-the-art detectors and observe that most hard false positives result from classification instead of localization and they have a large negative impact on the performance of object detectors. We conjecture three factors that lie behind hard false positives, and we confirm the conjecture with experiments that prove the following: (1) Shared feature representation is not optimal due to the mismatched goals of feature learning for classification and localization; (2) large receptive field for different scales leads to redundant context information for small objects; (3) multi-task learning helps, yet optimization of the multi-task loss may prove sub-optimal for individual tasks. We demonstrate the potential power of detector classification by a simple, effective, and widely applicable Decoupled Classification Refinement (DCR) network. In particular, DCR places a separate classification network in parallel with the localization network (base detector). With ROI pooling placed on the early stage of the classification network, we enforce an adaptive receptive field in DCR. During training, DCR samples hard false positives from the base detector and trains a strong classifier to refine classification results. During testing, DCR refines all boxes from the base detector. Experiments show competitive results on PASCAL VOC and COCO without any bells and whistles. Our codes are available at: https://github.com/bowenc0221/Decoupled-Classification-Refinement.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2020-02-28 without embargo terms","The student, Bowen Cheng, accepted the attached license on 2019-10-16 at 13:52.","The student, Bowen Cheng, submitted this Thesis for approval on 2019-10-16 at 13:59.","This Thesis was approved for publication on 2019-10-16 at 16:43.","DSpace SAF Submission Ingestion Package generated from Vireo submission #14500 on 2020-02-28 at 17:13:00","Made available in DSpace on 2020-03-02T21:58:05Z (GMT). No. of bitstreams: 3 CHENG-THESIS-2019.pdf: 3339997 bytes, checksum: 6aed1a4f984eb5c6e93565da85e10b3e (MD5) LICENSE.txt: 4208 bytes, checksum: 49dcaf80b3c39981f08301ac89ec45d7 (MD5) RightsLink Printable License.pdf: 115296 bytes, checksum: 08215077c5fa568e4fb59fa8ceec94e1 (MD5) Previous issue date: 2019-10-16"],"dc:format":["application/pdf"],"dc:identifier":["http://hdl.handle.net/2142/106175"],"dc:language":["en"],"dc:rights":["Copyright 2019 Bowen Cheng"],"dc:subject":["Object Detection","Deep Learning"],"dc:title":["Reducing false positives for object detection"],"dc:type":["text"],"thesis:degree_discipline":["Electrical & Computer Engr"],"thesis:degree_level":["Thesis"],"thesis:degree_name":["M.S."],"thesis:institution_name":["University of Illinois at Urbana-Champaign"]},"updated_at":"2026-07-22T22:24:45Z"}