Matches in SemOpenAlex for { <https://semopenalex.org/work/W4313135247> ?p ?o ?g. }
Showing items 1 to 81 of
81
with 100 items per page.
- W4313135247 abstract "Processing 3D data efficiently has always been a challenge. Spatial operations on large-scale point clouds, stored as sparse data, require extra cost. Attracted by the success of transformers, researchers are using multi-head attention for vision tasks. However, attention calculations in transformers come with quadratic complexity in the number of inputs and miss spatial intuition on sets like point clouds. We redesign set transformers in this work and incorporate them into a hierarchical framework for shape classification and part and scene segmentation. We propose our local attention unit, which captures features in a spatial neighborhood. We also compute efficient and dynamic global cross attentions by leveraging sampling and grouping at each iteration. Finally, to mitigate the non-heterogeneity of point clouds, we propose an efficient Multi-Scale Tokenization (MST), which extracts scale-invariant tokens for attention operations. The proposed hierarchical model achieves state-of-the-art shape classification in mean accuracy and yields results on par with the previous segmentation methods while requiring significantly fewer computations. Our proposed architecture predicts segmentation labels with around half the latency and parameter count of the previous most effi-cient method with comparable performance. The code is available at https://github.com/YigeWang-WHU/CloudAttention." @default.
- W4313135247 created "2023-01-06" @default.
- W4313135247 creator A5000247221 @default.
- W4313135247 creator A5041092666 @default.
- W4313135247 creator A5046896448 @default.
- W4313135247 creator A5067135033 @default.
- W4313135247 creator A5076333732 @default.
- W4313135247 date "2022-10-23" @default.
- W4313135247 modified "2023-10-16" @default.
- W4313135247 title "CloudAttention: Efficient Multi-Scale Attention Scheme For 3D Point Cloud Learning" @default.
- W4313135247 cites W2211722331 @default.
- W4313135247 cites W2460657278 @default.
- W4313135247 cites W2553307952 @default.
- W4313135247 cites W2788158258 @default.
- W4313135247 cites W2797997528 @default.
- W4313135247 cites W2963123724 @default.
- W4313135247 cites W2963226018 @default.
- W4313135247 cites W2963231572 @default.
- W4313135247 cites W2963517242 @default.
- W4313135247 cites W2963727135 @default.
- W4313135247 cites W3034314779 @default.
- W4313135247 cites W3122159272 @default.
- W4313135247 cites W3126711100 @default.
- W4313135247 cites W3136608016 @default.
- W4313135247 cites W3153465022 @default.
- W4313135247 cites W3205269569 @default.
- W4313135247 cites W3205586691 @default.
- W4313135247 cites W4312808629 @default.
- W4313135247 doi "https://doi.org/10.1109/iros47612.2022.9982276" @default.
- W4313135247 hasPublicationYear "2022" @default.
- W4313135247 type Work @default.
- W4313135247 citedByCount "0" @default.
- W4313135247 crossrefType "proceedings-article" @default.
- W4313135247 hasAuthorship W4313135247A5000247221 @default.
- W4313135247 hasAuthorship W4313135247A5041092666 @default.
- W4313135247 hasAuthorship W4313135247A5046896448 @default.
- W4313135247 hasAuthorship W4313135247A5067135033 @default.
- W4313135247 hasAuthorship W4313135247A5076333732 @default.
- W4313135247 hasBestOaLocation W43131352472 @default.
- W4313135247 hasConcept C111919701 @default.
- W4313135247 hasConcept C11413529 @default.
- W4313135247 hasConcept C119857082 @default.
- W4313135247 hasConcept C124101348 @default.
- W4313135247 hasConcept C131979681 @default.
- W4313135247 hasConcept C13670688 @default.
- W4313135247 hasConcept C154945302 @default.
- W4313135247 hasConcept C205711294 @default.
- W4313135247 hasConcept C41008148 @default.
- W4313135247 hasConcept C45374587 @default.
- W4313135247 hasConcept C55282118 @default.
- W4313135247 hasConcept C89600930 @default.
- W4313135247 hasConceptScore W4313135247C111919701 @default.
- W4313135247 hasConceptScore W4313135247C11413529 @default.
- W4313135247 hasConceptScore W4313135247C119857082 @default.
- W4313135247 hasConceptScore W4313135247C124101348 @default.
- W4313135247 hasConceptScore W4313135247C131979681 @default.
- W4313135247 hasConceptScore W4313135247C13670688 @default.
- W4313135247 hasConceptScore W4313135247C154945302 @default.
- W4313135247 hasConceptScore W4313135247C205711294 @default.
- W4313135247 hasConceptScore W4313135247C41008148 @default.
- W4313135247 hasConceptScore W4313135247C45374587 @default.
- W4313135247 hasConceptScore W4313135247C55282118 @default.
- W4313135247 hasConceptScore W4313135247C89600930 @default.
- W4313135247 hasLocation W43131352471 @default.
- W4313135247 hasLocation W43131352472 @default.
- W4313135247 hasLocation W43131352473 @default.
- W4313135247 hasOpenAccess W4313135247 @default.
- W4313135247 hasPrimaryLocation W43131352471 @default.
- W4313135247 hasRelatedWork W1990200345 @default.
- W4313135247 hasRelatedWork W2114282491 @default.
- W4313135247 hasRelatedWork W2562256921 @default.
- W4313135247 hasRelatedWork W2961085424 @default.
- W4313135247 hasRelatedWork W2971311575 @default.
- W4313135247 hasRelatedWork W3015465855 @default.
- W4313135247 hasRelatedWork W4214536195 @default.
- W4313135247 hasRelatedWork W4221144437 @default.
- W4313135247 hasRelatedWork W4306674287 @default.
- W4313135247 hasRelatedWork W4224009465 @default.
- W4313135247 isParatext "false" @default.
- W4313135247 isRetracted "false" @default.
- W4313135247 workType "article" @default.