Matches in SemOpenAlex for { <https://semopenalex.org/work/W4319300392> ?p ?o ?g. }
- W4319300392 abstract "Self-supervised monocular depth estimation (MDE) models universally suffer from the notorious edge-fattening issue. Triplet loss, as a widespread metric learning strategy, has largely succeeded in many computer vision applications. In this paper, we redesign the patch-based triplet loss in MDE to alleviate the ubiquitous edge-fattening issue. We show two drawbacks of the raw triplet loss in MDE and demonstrate our problem-driven redesigns. First, we present a min. operator based strategy applied to all negative samples, to prevent well-performing negatives sheltering the error of edge-fattening negatives. Second, we split the anchor-positive distance and anchor-negative distance from within the original triplet, which directly optimizes the positives without any mutual effect with the negatives. Extensive experiments show the combination of these two small redesigns can achieve unprecedented results: Our powerful and versatile triplet loss not only makes our model outperform all previous SoTA by a large margin, but also provides substantial performance boosts to a large number of existing models, while introducing no extra inference computation at all." @default.
- W4319300392 created "2023-02-07" @default.
- W4319300392 creator A5003642180 @default.
- W4319300392 creator A5016322200 @default.
- W4319300392 creator A5024911773 @default.
- W4319300392 creator A5032881192 @default.
- W4319300392 creator A5073310185 @default.
- W4319300392 creator A5076676389 @default.
- W4319300392 date "2023-01-01" @default.
- W4319300392 modified "2023-10-17" @default.
- W4319300392 title "Self-Supervised Monocular Depth Estimation: Solving the Edge-Fattening Problem" @default.
- W4319300392 cites W1975517671 @default.
- W4319300392 cites W2120657032 @default.
- W4319300392 cites W2133665775 @default.
- W4319300392 cites W2135442311 @default.
- W4319300392 cites W2138621090 @default.
- W4319300392 cites W2150066425 @default.
- W4319300392 cites W2157364932 @default.
- W4319300392 cites W2171590421 @default.
- W4319300392 cites W2520707372 @default.
- W4319300392 cites W2601564443 @default.
- W4319300392 cites W2609883120 @default.
- W4319300392 cites W2889002172 @default.
- W4319300392 cites W2894568782 @default.
- W4319300392 cites W2934279571 @default.
- W4319300392 cites W2963265330 @default.
- W4319300392 cites W2963412495 @default.
- W4319300392 cites W2963708168 @default.
- W4319300392 cites W2964076257 @default.
- W4319300392 cites W2964254867 @default.
- W4319300392 cites W2964968086 @default.
- W4319300392 cites W2967550825 @default.
- W4319300392 cites W2968529893 @default.
- W4319300392 cites W2981518351 @default.
- W4319300392 cites W2982102242 @default.
- W4319300392 cites W2985775862 @default.
- W4319300392 cites W3009185927 @default.
- W4319300392 cites W3034202663 @default.
- W4319300392 cites W3034428934 @default.
- W4319300392 cites W3034604951 @default.
- W4319300392 cites W3034953156 @default.
- W4319300392 cites W3043646556 @default.
- W4319300392 cites W3048510980 @default.
- W4319300392 cites W3060975791 @default.
- W4319300392 cites W3099206234 @default.
- W4319300392 cites W3118265840 @default.
- W4319300392 cites W3174581459 @default.
- W4319300392 cites W3175682855 @default.
- W4319300392 cites W3202016623 @default.
- W4319300392 cites W3206074207 @default.
- W4319300392 cites W4206545998 @default.
- W4319300392 doi "https://doi.org/10.1109/wacv56688.2023.00573" @default.
- W4319300392 hasPublicationYear "2023" @default.
- W4319300392 type Work @default.
- W4319300392 citedByCount "3" @default.
- W4319300392 countsByYear W43193003922022 @default.
- W4319300392 countsByYear W43193003922023 @default.
- W4319300392 crossrefType "proceedings-article" @default.
- W4319300392 hasAuthorship W4319300392A5003642180 @default.
- W4319300392 hasAuthorship W4319300392A5016322200 @default.
- W4319300392 hasAuthorship W4319300392A5024911773 @default.
- W4319300392 hasAuthorship W4319300392A5032881192 @default.
- W4319300392 hasAuthorship W4319300392A5073310185 @default.
- W4319300392 hasAuthorship W4319300392A5076676389 @default.
- W4319300392 hasBestOaLocation W43193003922 @default.
- W4319300392 hasConcept C112789634 @default.
- W4319300392 hasConcept C11413529 @default.
- W4319300392 hasConcept C119857082 @default.
- W4319300392 hasConcept C127413603 @default.
- W4319300392 hasConcept C154945302 @default.
- W4319300392 hasConcept C162307627 @default.
- W4319300392 hasConcept C176217482 @default.
- W4319300392 hasConcept C177264268 @default.
- W4319300392 hasConcept C199360897 @default.
- W4319300392 hasConcept C21547014 @default.
- W4319300392 hasConcept C2776214188 @default.
- W4319300392 hasConcept C2776760102 @default.
- W4319300392 hasConcept C31972630 @default.
- W4319300392 hasConcept C41008148 @default.
- W4319300392 hasConcept C45347329 @default.
- W4319300392 hasConcept C45374587 @default.
- W4319300392 hasConcept C50644808 @default.
- W4319300392 hasConcept C64869954 @default.
- W4319300392 hasConcept C65909025 @default.
- W4319300392 hasConcept C774472 @default.
- W4319300392 hasConceptScore W4319300392C112789634 @default.
- W4319300392 hasConceptScore W4319300392C11413529 @default.
- W4319300392 hasConceptScore W4319300392C119857082 @default.
- W4319300392 hasConceptScore W4319300392C127413603 @default.
- W4319300392 hasConceptScore W4319300392C154945302 @default.
- W4319300392 hasConceptScore W4319300392C162307627 @default.
- W4319300392 hasConceptScore W4319300392C176217482 @default.
- W4319300392 hasConceptScore W4319300392C177264268 @default.
- W4319300392 hasConceptScore W4319300392C199360897 @default.
- W4319300392 hasConceptScore W4319300392C21547014 @default.
- W4319300392 hasConceptScore W4319300392C2776214188 @default.
- W4319300392 hasConceptScore W4319300392C2776760102 @default.
- W4319300392 hasConceptScore W4319300392C31972630 @default.
- W4319300392 hasConceptScore W4319300392C41008148 @default.
- W4319300392 hasConceptScore W4319300392C45347329 @default.