Learning Open-World Object Proposals Without Learning to Classify

Cited 16 time in webofscience Cited 0 time in scopus
  • Hit : 359
  • Download : 0
DC FieldValueLanguage
dc.contributor.authorKim, Dahunko
dc.contributor.authorLin, Tsung-Yiko
dc.contributor.authorAngelova, Aneliako
dc.contributor.authorKweon, In Soko
dc.contributor.authorKuo, Weichengko
dc.date.accessioned2022-04-13T06:46:41Z-
dc.date.available2022-04-13T06:46:41Z-
dc.date.created2022-04-05-
dc.date.created2022-04-05-
dc.date.created2022-04-05-
dc.date.issued2022-04-
dc.identifier.citationIEEE ROBOTICS AND AUTOMATION LETTERS, v.7, no.2, pp.5453 - 5460-
dc.identifier.issn2377-3766-
dc.identifier.urihttp://hdl.handle.net/10203/292548-
dc.description.abstractObject proposals have become an integral pre-processing step of many vision pipelines including object detection, weakly supervised detection, object discovery, tracking, etc. Compared to the learning-free methods, learning-based proposals have become popular recently due to the growing interest in object detection. The common paradigm is to learn object proposals from data labeled with a set of object regions and their corresponding categories. However, this approach often struggles with novel objects in the open world that are absent in the training set. In this letter, we identify that the problem is that the binary classifiers in existing proposal methods tend to overfit to the training categories. Therefore, we propose a classification-free Object Localization Network (OLN) which estimates the objectness of each region purely by how well the location and shape of a region overlap with any ground-truth object (e.g., centerness and IoU). This strategy learns generalizable objectness and outperforms existing proposals on cross-category generalization on COCO. We further explore more challenging cross-dataset generalization onto RoboNet and EpicKitchens dataset, and long-tail detection on LVIS dataset. We demonstrate clear improvement over the state-of-the-art object detectors and object proposers. The code is publicly available at https://github.com/mcahny/object_localization_network.-
dc.languageEnglish-
dc.publisherIEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC-
dc.titleLearning Open-World Object Proposals Without Learning to Classify-
dc.typeArticle-
dc.identifier.wosid000770589900007-
dc.identifier.scopusid2-s2.0-85124195174-
dc.type.rimsART-
dc.citation.volume7-
dc.citation.issue2-
dc.citation.beginningpage5453-
dc.citation.endingpage5460-
dc.citation.publicationnameIEEE ROBOTICS AND AUTOMATION LETTERS-
dc.identifier.doi10.1109/LRA.2022.3146922-
dc.contributor.localauthorKweon, In So-
dc.contributor.nonIdAuthorLin, Tsung-Yi-
dc.contributor.nonIdAuthorAngelova, Anelia-
dc.contributor.nonIdAuthorKuo, Weicheng-
dc.description.isOpenAccessN-
dc.type.journalArticleArticle-
dc.subject.keywordAuthorOpen-world detection/segmentation-
dc.subject.keywordPlusREGION PROPOSAL-
Appears in Collection
EE-Journal Papers(저널논문)
Files in This Item
There are no files associated with this item.
This item is cited by other documents in WoS
⊙ Detail Information in WoSⓡ Click to see webofscience_button
⊙ Cited 16 items in WoS Click to see citing articles in records_button

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0