Matches in SemOpenAlex for { <https://semopenalex.org/work/W3175457067> ?p ?o ?g. }
- W3175457067 abstract "Inverse rendering is the problem of decomposing an image into its intrinsic components, i.e. albedo, normal and lighting. To solve this ill-posed problem from single image, state-of-the-art methods in shape from shading mostly resort to supervised training on all the components on either synthetic or real datasets. Here, we propose a new self-supervised training paradigm that 1) reduces the need for full supervision on the decomposition task and 2) takes into account the relighting task. We introduce new self-supervised loss terms that leverage the consistencies between multi-lit images (images of the same scene under different illuminations). Our approach is applicable to multi-lit datasets. We apply our training approach in two settings: 1) train on a mixture of synthetic and real data, 2) train on real datasets with limited supervision. We show-case the effectiveness of our training paradigm on both intrinsic decomposition and relighting and demonstrate how the model struggles in both tasks without the self-supervised loss terms in limited supervision settings. We provide results of comprehensive experiments on SfSNet, CelebA and Photoface datasets and verify the performance of our approach on images in the wild." @default.
- W3175457067 created "2021-07-05" @default.
- W3175457067 creator A5014657145 @default.
- W3175457067 creator A5027156022 @default.
- W3175457067 creator A5031378199 @default.
- W3175457067 creator A5031811971 @default.
- W3175457067 date "2021-06-22" @default.
- W3175457067 modified "2023-09-27" @default.
- W3175457067 title "Joint Learning of Portrait Intrinsic Decomposition and Relighting." @default.
- W3175457067 cites W1691728462 @default.
- W3175457067 cites W1834627138 @default.
- W3175457067 cites W1836465849 @default.
- W3175457067 cites W1890563702 @default.
- W3175457067 cites W1921523184 @default.
- W3175457067 cites W2022553599 @default.
- W3175457067 cites W2105649179 @default.
- W3175457067 cites W2106395586 @default.
- W3175457067 cites W2115252128 @default.
- W3175457067 cites W2116919352 @default.
- W3175457067 cites W2121314241 @default.
- W3175457067 cites W2125389028 @default.
- W3175457067 cites W2138127543 @default.
- W3175457067 cites W2164847484 @default.
- W3175457067 cites W2193045528 @default.
- W3175457067 cites W2235136907 @default.
- W3175457067 cites W2237250383 @default.
- W3175457067 cites W2342277278 @default.
- W3175457067 cites W2567309586 @default.
- W3175457067 cites W2604672468 @default.
- W3175457067 cites W2607170299 @default.
- W3175457067 cites W2738763667 @default.
- W3175457067 cites W2767225062 @default.
- W3175457067 cites W2810337759 @default.
- W3175457067 cites W2885621364 @default.
- W3175457067 cites W2890382763 @default.
- W3175457067 cites W2895013824 @default.
- W3175457067 cites W2903365351 @default.
- W3175457067 cites W2953952939 @default.
- W3175457067 cites W2954058873 @default.
- W3175457067 cites W2960236945 @default.
- W3175457067 cites W2962780596 @default.
- W3175457067 cites W2963073614 @default.
- W3175457067 cites W2963390109 @default.
- W3175457067 cites W2963498646 @default.
- W3175457067 cites W2963644875 @default.
- W3175457067 cites W2963683323 @default.
- W3175457067 cites W2963917315 @default.
- W3175457067 cites W2964321964 @default.
- W3175457067 cites W3007712693 @default.
- W3175457067 cites W3009953436 @default.
- W3175457067 cites W3034514298 @default.
- W3175457067 cites W3034868405 @default.
- W3175457067 cites W3121857126 @default.
- W3175457067 hasPublicationYear "2021" @default.
- W3175457067 type Work @default.
- W3175457067 sameAs 3175457067 @default.
- W3175457067 citedByCount "0" @default.
- W3175457067 crossrefType "posted-content" @default.
- W3175457067 hasAuthorship W3175457067A5014657145 @default.
- W3175457067 hasAuthorship W3175457067A5027156022 @default.
- W3175457067 hasAuthorship W3175457067A5031378199 @default.
- W3175457067 hasAuthorship W3175457067A5031811971 @default.
- W3175457067 hasConcept C119857082 @default.
- W3175457067 hasConcept C124681953 @default.
- W3175457067 hasConcept C153083717 @default.
- W3175457067 hasConcept C153180895 @default.
- W3175457067 hasConcept C154945302 @default.
- W3175457067 hasConcept C162324750 @default.
- W3175457067 hasConcept C187736073 @default.
- W3175457067 hasConcept C18903297 @default.
- W3175457067 hasConcept C205711294 @default.
- W3175457067 hasConcept C2780451532 @default.
- W3175457067 hasConcept C31972630 @default.
- W3175457067 hasConcept C41008148 @default.
- W3175457067 hasConcept C51632099 @default.
- W3175457067 hasConcept C86803240 @default.
- W3175457067 hasConceptScore W3175457067C119857082 @default.
- W3175457067 hasConceptScore W3175457067C124681953 @default.
- W3175457067 hasConceptScore W3175457067C153083717 @default.
- W3175457067 hasConceptScore W3175457067C153180895 @default.
- W3175457067 hasConceptScore W3175457067C154945302 @default.
- W3175457067 hasConceptScore W3175457067C162324750 @default.
- W3175457067 hasConceptScore W3175457067C187736073 @default.
- W3175457067 hasConceptScore W3175457067C18903297 @default.
- W3175457067 hasConceptScore W3175457067C205711294 @default.
- W3175457067 hasConceptScore W3175457067C2780451532 @default.
- W3175457067 hasConceptScore W3175457067C31972630 @default.
- W3175457067 hasConceptScore W3175457067C41008148 @default.
- W3175457067 hasConceptScore W3175457067C51632099 @default.
- W3175457067 hasConceptScore W3175457067C86803240 @default.
- W3175457067 hasLocation W31754570671 @default.
- W3175457067 hasOpenAccess W3175457067 @default.
- W3175457067 hasPrimaryLocation W31754570671 @default.
- W3175457067 hasRelatedWork W2567101557 @default.
- W3175457067 hasRelatedWork W2792892164 @default.
- W3175457067 hasRelatedWork W2795464554 @default.
- W3175457067 hasRelatedWork W2883993491 @default.
- W3175457067 hasRelatedWork W2888277922 @default.
- W3175457067 hasRelatedWork W2895013824 @default.
- W3175457067 hasRelatedWork W2951851780 @default.