Matches in SemOpenAlex for { <https://semopenalex.org/work/W2949892238> ?p ?o ?g. }
- W2949892238 abstract "Convolutional neural networks (CNNs) have achieved great successes in many computer vision problems. Unlike existing works that designed CNN architectures to improve performance on a single task of a single domain and not generalizable, we present IBN-Net, a novel convolutional architecture, which remarkably enhances a CNN's modeling ability on one domain (e.g. Cityscapes) as well as its generalization capacity on another domain (e.g. GTA5) without finetuning. IBN-Net carefully integrates Instance Normalization (IN) and Batch Normalization (BN) as building blocks, and can be wrapped into many advanced deep networks to improve their performances. This work has three key contributions. (1) By delving into IN and BN, we disclose that IN learns features that are invariant to appearance changes, such as colors, styles, and virtuality/reality, while BN is essential for preserving content related information. (2) IBN-Net can be applied to many advanced deep architectures, such as DenseNet, ResNet, ResNeXt, and SENet, and consistently improve their performance without increasing computational cost. (3) When applying the trained networks to new domains, e.g. from GTA5 to Cityscapes, IBN-Net achieves comparable improvements as domain adaptation methods, even without using data from the target domain. With IBN-Net, we won the 1st place on the WAD 2018 Challenge Drivable Area track, with an mIoU of 86.18%." @default.
- W2949892238 created "2019-06-27" @default.
- W2949892238 creator A5036257681 @default.
- W2949892238 creator A5052549072 @default.
- W2949892238 creator A5066603429 @default.
- W2949892238 creator A5075158591 @default.
- W2949892238 date "2018-07-25" @default.
- W2949892238 modified "2023-10-16" @default.
- W2949892238 title "Two at Once: Enhancing Learning and Generalization Capacities via IBN-Net" @default.
- W2949892238 cites W1565327149 @default.
- W2949892238 cites W1836465849 @default.
- W2949892238 cites W1852255964 @default.
- W2949892238 cites W1920962657 @default.
- W2949892238 cites W2095705004 @default.
- W2949892238 cites W2097117768 @default.
- W2949892238 cites W2108598243 @default.
- W2949892238 cites W2155858138 @default.
- W2949892238 cites W2159291411 @default.
- W2949892238 cites W2163605009 @default.
- W2949892238 cites W2194775991 @default.
- W2949892238 cites W2299668505 @default.
- W2949892238 cites W2302255633 @default.
- W2949892238 cites W2340897893 @default.
- W2949892238 cites W2412782625 @default.
- W2949892238 cites W2487365028 @default.
- W2949892238 cites W2545656684 @default.
- W2949892238 cites W2549139847 @default.
- W2949892238 cites W2562192638 @default.
- W2949892238 cites W2572730214 @default.
- W2949892238 cites W2593768305 @default.
- W2949892238 cites W2601564443 @default.
- W2949892238 cites W2603777577 @default.
- W2949892238 cites W2613718673 @default.
- W2949892238 cites W2769672616 @default.
- W2949892238 cites W2962793481 @default.
- W2949892238 cites W2962808524 @default.
- W2949892238 cites W2963420686 @default.
- W2949892238 cites W2963446712 @default.
- W2949892238 cites W2963611454 @default.
- W2949892238 cites W2964288524 @default.
- W2949892238 doi "https://doi.org/10.48550/arxiv.1807.09441" @default.
- W2949892238 hasPublicationYear "2018" @default.
- W2949892238 type Work @default.
- W2949892238 sameAs 2949892238 @default.
- W2949892238 citedByCount "0" @default.
- W2949892238 crossrefType "posted-content" @default.
- W2949892238 hasAuthorship W2949892238A5036257681 @default.
- W2949892238 hasAuthorship W2949892238A5052549072 @default.
- W2949892238 hasAuthorship W2949892238A5066603429 @default.
- W2949892238 hasAuthorship W2949892238A5075158591 @default.
- W2949892238 hasBestOaLocation W29498922381 @default.
- W2949892238 hasConcept C108583219 @default.
- W2949892238 hasConcept C119857082 @default.
- W2949892238 hasConcept C123657996 @default.
- W2949892238 hasConcept C134306372 @default.
- W2949892238 hasConcept C136886441 @default.
- W2949892238 hasConcept C142362112 @default.
- W2949892238 hasConcept C144024400 @default.
- W2949892238 hasConcept C153349607 @default.
- W2949892238 hasConcept C154945302 @default.
- W2949892238 hasConcept C177148314 @default.
- W2949892238 hasConcept C19165224 @default.
- W2949892238 hasConcept C2776434776 @default.
- W2949892238 hasConcept C33923547 @default.
- W2949892238 hasConcept C41008148 @default.
- W2949892238 hasConcept C81363708 @default.
- W2949892238 hasConcept C95623464 @default.
- W2949892238 hasConceptScore W2949892238C108583219 @default.
- W2949892238 hasConceptScore W2949892238C119857082 @default.
- W2949892238 hasConceptScore W2949892238C123657996 @default.
- W2949892238 hasConceptScore W2949892238C134306372 @default.
- W2949892238 hasConceptScore W2949892238C136886441 @default.
- W2949892238 hasConceptScore W2949892238C142362112 @default.
- W2949892238 hasConceptScore W2949892238C144024400 @default.
- W2949892238 hasConceptScore W2949892238C153349607 @default.
- W2949892238 hasConceptScore W2949892238C154945302 @default.
- W2949892238 hasConceptScore W2949892238C177148314 @default.
- W2949892238 hasConceptScore W2949892238C19165224 @default.
- W2949892238 hasConceptScore W2949892238C2776434776 @default.
- W2949892238 hasConceptScore W2949892238C33923547 @default.
- W2949892238 hasConceptScore W2949892238C41008148 @default.
- W2949892238 hasConceptScore W2949892238C81363708 @default.
- W2949892238 hasConceptScore W2949892238C95623464 @default.
- W2949892238 hasLocation W29498922381 @default.
- W2949892238 hasLocation W29498922382 @default.
- W2949892238 hasOpenAccess W2949892238 @default.
- W2949892238 hasPrimaryLocation W29498922381 @default.
- W2949892238 hasRelatedWork W2337926734 @default.
- W2949892238 hasRelatedWork W2731899572 @default.
- W2949892238 hasRelatedWork W3133861977 @default.
- W2949892238 hasRelatedWork W4200173597 @default.
- W2949892238 hasRelatedWork W4312417841 @default.
- W2949892238 hasRelatedWork W4320802194 @default.
- W2949892238 hasRelatedWork W4321369474 @default.
- W2949892238 hasRelatedWork W4366224123 @default.
- W2949892238 hasRelatedWork W4381487685 @default.
- W2949892238 hasRelatedWork W4381832759 @default.
- W2949892238 isParatext "false" @default.
- W2949892238 isRetracted "false" @default.
- W2949892238 magId "2949892238" @default.