Matches in SemOpenAlex for { <https://semopenalex.org/work/W4367060909> ?p ?o ?g. }
Showing items 1 to 61 of
61
with 100 items per page.
- W4367060909 abstract "Monocular depth estimation is very challenging because clues to the exact depth are incomplete in a single RGB image. To overcome the limitation, deep neural networks rely on various visual hints such as size, shade, and texture extracted from RGB information. However, we observe that if such hints are overly exploited, the network can be biased on RGB information without considering the comprehensive view. We propose a novel depth estimation model named RElative Depth Transformer (RED-T) that uses relative depth as guidance in self-attention. Specifically, the model assigns high attention weights to pixels of close depth and low attention weights to pixels of distant depth. As a result, the features of similar depth can become more likely to each other and thus less prone to misused visual hints. We show that the proposed model achieves competitive results in monocular depth estimation benchmarks and is less biased to RGB information. In addition, we propose a novel monocular depth estimation benchmark that limits the observable depth range during training in order to evaluate the robustness of the model for unseen depths." @default.
- W4367060909 created "2023-04-27" @default.
- W4367060909 creator A5011150122 @default.
- W4367060909 creator A5041792310 @default.
- W4367060909 creator A5064051041 @default.
- W4367060909 creator A5076075267 @default.
- W4367060909 date "2023-04-25" @default.
- W4367060909 modified "2023-09-24" @default.
- W4367060909 title "Depth-Relative Self Attention for Monocular Depth Estimation" @default.
- W4367060909 doi "https://doi.org/10.48550/arxiv.2304.12849" @default.
- W4367060909 hasPublicationYear "2023" @default.
- W4367060909 type Work @default.
- W4367060909 citedByCount "0" @default.
- W4367060909 crossrefType "posted-content" @default.
- W4367060909 hasAuthorship W4367060909A5011150122 @default.
- W4367060909 hasAuthorship W4367060909A5041792310 @default.
- W4367060909 hasAuthorship W4367060909A5064051041 @default.
- W4367060909 hasAuthorship W4367060909A5076075267 @default.
- W4367060909 hasBestOaLocation W43670609091 @default.
- W4367060909 hasConcept C104317684 @default.
- W4367060909 hasConcept C127313418 @default.
- W4367060909 hasConcept C13280743 @default.
- W4367060909 hasConcept C154945302 @default.
- W4367060909 hasConcept C160633673 @default.
- W4367060909 hasConcept C185592680 @default.
- W4367060909 hasConcept C185798385 @default.
- W4367060909 hasConcept C31972630 @default.
- W4367060909 hasConcept C41008148 @default.
- W4367060909 hasConcept C55493867 @default.
- W4367060909 hasConcept C63479239 @default.
- W4367060909 hasConcept C65909025 @default.
- W4367060909 hasConcept C82990744 @default.
- W4367060909 hasConceptScore W4367060909C104317684 @default.
- W4367060909 hasConceptScore W4367060909C127313418 @default.
- W4367060909 hasConceptScore W4367060909C13280743 @default.
- W4367060909 hasConceptScore W4367060909C154945302 @default.
- W4367060909 hasConceptScore W4367060909C160633673 @default.
- W4367060909 hasConceptScore W4367060909C185592680 @default.
- W4367060909 hasConceptScore W4367060909C185798385 @default.
- W4367060909 hasConceptScore W4367060909C31972630 @default.
- W4367060909 hasConceptScore W4367060909C41008148 @default.
- W4367060909 hasConceptScore W4367060909C55493867 @default.
- W4367060909 hasConceptScore W4367060909C63479239 @default.
- W4367060909 hasConceptScore W4367060909C65909025 @default.
- W4367060909 hasConceptScore W4367060909C82990744 @default.
- W4367060909 hasLocation W43670609091 @default.
- W4367060909 hasOpenAccess W4367060909 @default.
- W4367060909 hasPrimaryLocation W43670609091 @default.
- W4367060909 hasRelatedWork W2139599928 @default.
- W4367060909 hasRelatedWork W2149848058 @default.
- W4367060909 hasRelatedWork W2551317114 @default.
- W4367060909 hasRelatedWork W2767823485 @default.
- W4367060909 hasRelatedWork W2794771666 @default.
- W4367060909 hasRelatedWork W2921707373 @default.
- W4367060909 hasRelatedWork W3010888991 @default.
- W4367060909 hasRelatedWork W4213228110 @default.
- W4367060909 hasRelatedWork W4321512589 @default.
- W4367060909 hasRelatedWork W846293480 @default.
- W4367060909 isParatext "false" @default.
- W4367060909 isRetracted "false" @default.
- W4367060909 workType "article" @default.