Matches in SemOpenAlex for { <https://semopenalex.org/work/W3017788695> ?p ?o ?g. }
Showing items 1 to 83 of
83
with 100 items per page.
- W3017788695 endingPage "381" @default.
- W3017788695 startingPage "365" @default.
- W3017788695 abstract "We propose a novel weakly supervised learning segmentation based on several global constraints derived from box annotations. Particularly, we leverage a classical tightness prior to a deep learning setting via imposing a set of constraints on the network outputs. Such a powerful topological prior prevents solutions from excessive shrinking by enforcing any horizontal or vertical line within the bounding box to contain, at least, one pixel of the foreground region. Furthermore, we integrate our deep tightness prior with a global background emptiness constraint, guiding training with information outside the bounding box. We demonstrate experimentally that such a global constraint is much more powerful than standard cross-entropy for the background class. Our optimization problem is challenging as it takes the form of a large set of inequality constraints on the outputs of deep networks. We solve it with sequence of unconstrained losses based on a recent powerful extension of the log-barrier method, which is well-known in the context of interior-point methods. This accommodates standard stochastic gradient descent (SGD) for training deep networks, while avoiding computationally expensive and unstable Lagrangian dual steps and projections. Extensive experiments over two different public data sets and applications (prostate and brain lesions) demonstrate that the synergy between our global tightness and emptiness priors yield very competitive performances, approaching full supervision and outperforming significantly DeepCut. Furthermore, our approach removes the need for computationally expensive proposal generation. Our code is shared anonymously." @default.
- W3017788695 created "2020-05-01" @default.
- W3017788695 creator A5004770604 @default.
- W3017788695 creator A5006937759 @default.
- W3017788695 creator A5053292735 @default.
- W3017788695 creator A5055299862 @default.
- W3017788695 creator A5057382610 @default.
- W3017788695 date "2020-01-25" @default.
- W3017788695 modified "2023-09-26" @default.
- W3017788695 title "Bounding boxes for weakly supervised segmentation: Global constraints get close to full supervision" @default.
- W3017788695 hasPublicationYear "2020" @default.
- W3017788695 type Work @default.
- W3017788695 sameAs 3017788695 @default.
- W3017788695 citedByCount "13" @default.
- W3017788695 countsByYear W30177886952020 @default.
- W3017788695 countsByYear W30177886952021 @default.
- W3017788695 crossrefType "journal-article" @default.
- W3017788695 hasAuthorship W3017788695A5004770604 @default.
- W3017788695 hasAuthorship W3017788695A5006937759 @default.
- W3017788695 hasAuthorship W3017788695A5053292735 @default.
- W3017788695 hasAuthorship W3017788695A5055299862 @default.
- W3017788695 hasAuthorship W3017788695A5057382610 @default.
- W3017788695 hasConcept C108583219 @default.
- W3017788695 hasConcept C11413529 @default.
- W3017788695 hasConcept C115961682 @default.
- W3017788695 hasConcept C119857082 @default.
- W3017788695 hasConcept C126255220 @default.
- W3017788695 hasConcept C147037132 @default.
- W3017788695 hasConcept C151730666 @default.
- W3017788695 hasConcept C153083717 @default.
- W3017788695 hasConcept C154945302 @default.
- W3017788695 hasConcept C2779343474 @default.
- W3017788695 hasConcept C33923547 @default.
- W3017788695 hasConcept C41008148 @default.
- W3017788695 hasConcept C63584917 @default.
- W3017788695 hasConcept C774472 @default.
- W3017788695 hasConcept C86803240 @default.
- W3017788695 hasConcept C89600930 @default.
- W3017788695 hasConceptScore W3017788695C108583219 @default.
- W3017788695 hasConceptScore W3017788695C11413529 @default.
- W3017788695 hasConceptScore W3017788695C115961682 @default.
- W3017788695 hasConceptScore W3017788695C119857082 @default.
- W3017788695 hasConceptScore W3017788695C126255220 @default.
- W3017788695 hasConceptScore W3017788695C147037132 @default.
- W3017788695 hasConceptScore W3017788695C151730666 @default.
- W3017788695 hasConceptScore W3017788695C153083717 @default.
- W3017788695 hasConceptScore W3017788695C154945302 @default.
- W3017788695 hasConceptScore W3017788695C2779343474 @default.
- W3017788695 hasConceptScore W3017788695C33923547 @default.
- W3017788695 hasConceptScore W3017788695C41008148 @default.
- W3017788695 hasConceptScore W3017788695C63584917 @default.
- W3017788695 hasConceptScore W3017788695C774472 @default.
- W3017788695 hasConceptScore W3017788695C86803240 @default.
- W3017788695 hasConceptScore W3017788695C89600930 @default.
- W3017788695 hasLocation W30177886951 @default.
- W3017788695 hasOpenAccess W3017788695 @default.
- W3017788695 hasPrimaryLocation W30177886951 @default.
- W3017788695 hasRelatedWork W1495267108 @default.
- W3017788695 hasRelatedWork W1686810756 @default.
- W3017788695 hasRelatedWork W1783315696 @default.
- W3017788695 hasRelatedWork W1901129140 @default.
- W3017788695 hasRelatedWork W1903029394 @default.
- W3017788695 hasRelatedWork W2124351162 @default.
- W3017788695 hasRelatedWork W2194775991 @default.
- W3017788695 hasRelatedWork W2221898772 @default.
- W3017788695 hasRelatedWork W2295107390 @default.
- W3017788695 hasRelatedWork W2337429362 @default.
- W3017788695 hasRelatedWork W2396622801 @default.
- W3017788695 hasRelatedWork W2412782625 @default.
- W3017788695 hasRelatedWork W2552414813 @default.
- W3017788695 hasRelatedWork W2567599812 @default.
- W3017788695 hasRelatedWork W2799738340 @default.
- W3017788695 hasRelatedWork W2962914239 @default.
- W3017788695 hasRelatedWork W2963687373 @default.
- W3017788695 hasRelatedWork W2979522411 @default.
- W3017788695 hasRelatedWork W3017003173 @default.
- W3017788695 hasRelatedWork W611457968 @default.
- W3017788695 isParatext "false" @default.
- W3017788695 isRetracted "false" @default.
- W3017788695 magId "3017788695" @default.
- W3017788695 workType "article" @default.