Distill-to-Label: Weakly Supervised Instance Labeling Using Knowledge Distillation
Distill-to-Label: Weakly Supervised Instance Labeling Using Knowledge Distillation
About this item
Full title
Author / Creator
Publisher
Ithaca: Cornell University Library, arXiv.org
Journal title
Language
English
Formats
Publication information
Publisher
Ithaca: Cornell University Library, arXiv.org
Subjects
More information
Scope and Contents
Contents
Weakly supervised instance labeling using only image-level labels, in lieu of expensive fine-grained pixel annotations, is crucial in several applications including medical image analysis. In contrast to conventional instance segmentation scenarios in computer vision, the problems that we consider are characterized by a small number of training ima...
Alternative Titles
Full title
Distill-to-Label: Weakly Supervised Instance Labeling Using Knowledge Distillation
Authors, Artists and Contributors
Author / Creator
Identifiers
Primary Identifiers
Record Identifier
TN_cdi_proquest_journals_2267319062
Permalink
https://devfeature-collection.sl.nsw.gov.au/record/TN_cdi_proquest_journals_2267319062
Other Identifiers
E-ISSN
2331-8422
