Matches in SemOpenAlex for { <https://semopenalex.org/work/W4313641664> ?p ?o ?g. }
Showing items 1 to 88 of
88
with 100 items per page.
- W4313641664 endingPage "13" @default.
- W4313641664 startingPage "1" @default.
- W4313641664 abstract "3D human reconstruction technology is widely used in VR/AR content creation, image/video editing, telepresence, and virtual dressing. Compared with existing methods of 3D human representations, the pixel-aligned implicit function (PIFu) representation is more memory efficient and solves the feature ambiguity problem inherent to single-view reconstruction. However, it does not consider global geometric consistency, which may lead to unnatural body shapes in reconstructions. We propose TransPIFu, a novel implicit representation combining Transformer and PIFu representations to achieve high-quality depth perception of the clothed human for reconstructing 3D clothed humans from single-view images. We explore the geometric correlation between 3D feature blocks with Transformer, which can help us infer the geometric structure of invisible regions under different views and ensure the global consistency of the reconstruction as a whole. We effectively predict the local details from visible and invisible regions by combining the predicted front/back normal maps of the clothed human obtained from RGB images. Our reconstruction results have less shape distortion and better surface details than existing methods and are more efficient compared to the two-stage reconstruction methods. Experiments on the DeepHuman dataset, the BUFF dataset and the THUman2.0 dataset demonstrate that TransPIFu has better reconstruction results both globally and locally." @default.
- W4313641664 created "2023-01-07" @default.
- W4313641664 creator A5018323754 @default.
- W4313641664 creator A5028149360 @default.
- W4313641664 creator A5063584181 @default.
- W4313641664 date "2023-04-01" @default.
- W4313641664 modified "2023-10-17" @default.
- W4313641664 title "TransPIFu: Combining Transformer and Pixel-Aligned Implicit Function for Single-view Clothed Human Reconstruction" @default.
- W4313641664 cites W1967554269 @default.
- W4313641664 cites W2082211440 @default.
- W4313641664 cites W2596210417 @default.
- W4313641664 cites W2797515701 @default.
- W4313641664 cites W2962849139 @default.
- W4313641664 cites W2962921964 @default.
- W4313641664 cites W2963627347 @default.
- W4313641664 cites W2963800363 @default.
- W4313641664 cites W2963926543 @default.
- W4313641664 cites W2964219767 @default.
- W4313641664 cites W2968257580 @default.
- W4313641664 cites W2978956737 @default.
- W4313641664 cites W2981978060 @default.
- W4313641664 cites W2991621301 @default.
- W4313641664 cites W3034502973 @default.
- W4313641664 cites W3035291735 @default.
- W4313641664 cites W3035492592 @default.
- W4313641664 cites W3035507572 @default.
- W4313641664 cites W3041416670 @default.
- W4313641664 cites W3174025609 @default.
- W4313641664 cites W3175408382 @default.
- W4313641664 cites W3176762345 @default.
- W4313641664 cites W3202432020 @default.
- W4313641664 cites W4212875960 @default.
- W4313641664 cites W4214655709 @default.
- W4313641664 cites W4214755140 @default.
- W4313641664 cites W4226160668 @default.
- W4313641664 cites W4233857083 @default.
- W4313641664 cites W4312793185 @default.
- W4313641664 cites W4313175219 @default.
- W4313641664 doi "https://doi.org/10.1016/j.cag.2022.12.009" @default.
- W4313641664 hasPublicationYear "2023" @default.
- W4313641664 type Work @default.
- W4313641664 citedByCount "2" @default.
- W4313641664 countsByYear W43136416642023 @default.
- W4313641664 crossrefType "journal-article" @default.
- W4313641664 hasAuthorship W4313641664A5018323754 @default.
- W4313641664 hasAuthorship W4313641664A5028149360 @default.
- W4313641664 hasAuthorship W4313641664A5063584181 @default.
- W4313641664 hasConcept C109950114 @default.
- W4313641664 hasConcept C121332964 @default.
- W4313641664 hasConcept C126057942 @default.
- W4313641664 hasConcept C141379421 @default.
- W4313641664 hasConcept C154945302 @default.
- W4313641664 hasConcept C160633673 @default.
- W4313641664 hasConcept C165801399 @default.
- W4313641664 hasConcept C31972630 @default.
- W4313641664 hasConcept C41008148 @default.
- W4313641664 hasConcept C62520636 @default.
- W4313641664 hasConcept C66322947 @default.
- W4313641664 hasConceptScore W4313641664C109950114 @default.
- W4313641664 hasConceptScore W4313641664C121332964 @default.
- W4313641664 hasConceptScore W4313641664C126057942 @default.
- W4313641664 hasConceptScore W4313641664C141379421 @default.
- W4313641664 hasConceptScore W4313641664C154945302 @default.
- W4313641664 hasConceptScore W4313641664C160633673 @default.
- W4313641664 hasConceptScore W4313641664C165801399 @default.
- W4313641664 hasConceptScore W4313641664C31972630 @default.
- W4313641664 hasConceptScore W4313641664C41008148 @default.
- W4313641664 hasConceptScore W4313641664C62520636 @default.
- W4313641664 hasConceptScore W4313641664C66322947 @default.
- W4313641664 hasLocation W43136416641 @default.
- W4313641664 hasOpenAccess W4313641664 @default.
- W4313641664 hasPrimaryLocation W43136416641 @default.
- W4313641664 hasRelatedWork W1536158975 @default.
- W4313641664 hasRelatedWork W1983607852 @default.
- W4313641664 hasRelatedWork W2061445396 @default.
- W4313641664 hasRelatedWork W2073129023 @default.
- W4313641664 hasRelatedWork W2157670837 @default.
- W4313641664 hasRelatedWork W2288745733 @default.
- W4313641664 hasRelatedWork W2374120702 @default.
- W4313641664 hasRelatedWork W2383936069 @default.
- W4313641664 hasRelatedWork W2393788985 @default.
- W4313641664 hasRelatedWork W2631594184 @default.
- W4313641664 hasVolume "111" @default.
- W4313641664 isParatext "false" @default.
- W4313641664 isRetracted "false" @default.
- W4313641664 workType "article" @default.