Matches in SemOpenAlex for { <https://semopenalex.org/work/W4310921857> ?p ?o ?g. }
Showing items 1 to 70 of
70
with 100 items per page.
- W4310921857 abstract "Recent advances in pixel-level tasks (e.g., segmentation) illustrate the benefit of long-range interactions between aggregated region-based representations that can enhance local features. However, such pixel-to-region associations and the resulting representation, which often take the form of attention, cannot model the underlying semantic structure of the scene (e.g., individual objects and, by extension, their interactions). In this work, we take a step toward addressing this limitation. Specifically, we propose an architecture where we learn to project image features into latent region representations and perform global reasoning across them, using a transformer, to produce contextualized and scene-consistent representations that are then fused with original pixel-level features. Our design enables the latent regions to represent semantically meaningful concepts, by ensuring that activated regions are spatially disjoint and unions of such regions correspond to connected object segments. The resulting semantic global reasoning (SGR) is end-to-end trainable and can be combined with any semantic segmentation framework and backbone. Combining SGR with DeepLabV3 results in a semantic segmentation performance that is competitive to the state-of-the-art, while resulting in more semantically interpretable and diverse region representations, which we show can effectively transfer to detection and instance segmentation. Further, we propose a new metric that allows us to measure the semantics of representations at both the object class and instance level." @default.
- W4310921857 created "2022-12-21" @default.
- W4310921857 creator A5007777373 @default.
- W4310921857 creator A5053011888 @default.
- W4310921857 creator A5088358011 @default.
- W4310921857 date "2022-12-06" @default.
- W4310921857 modified "2023-10-16" @default.
- W4310921857 title "Semantically Enhanced Global Reasoning for Semantic Segmentation" @default.
- W4310921857 doi "https://doi.org/10.48550/arxiv.2212.03338" @default.
- W4310921857 hasPublicationYear "2022" @default.
- W4310921857 type Work @default.
- W4310921857 citedByCount "0" @default.
- W4310921857 crossrefType "posted-content" @default.
- W4310921857 hasAuthorship W4310921857A5007777373 @default.
- W4310921857 hasAuthorship W4310921857A5053011888 @default.
- W4310921857 hasAuthorship W4310921857A5088358011 @default.
- W4310921857 hasBestOaLocation W43109218571 @default.
- W4310921857 hasConcept C119857082 @default.
- W4310921857 hasConcept C121332964 @default.
- W4310921857 hasConcept C153180895 @default.
- W4310921857 hasConcept C154945302 @default.
- W4310921857 hasConcept C160633673 @default.
- W4310921857 hasConcept C165801399 @default.
- W4310921857 hasConcept C17744445 @default.
- W4310921857 hasConcept C184337299 @default.
- W4310921857 hasConcept C199360897 @default.
- W4310921857 hasConcept C199539241 @default.
- W4310921857 hasConcept C204321447 @default.
- W4310921857 hasConcept C2776359362 @default.
- W4310921857 hasConcept C2781238097 @default.
- W4310921857 hasConcept C41008148 @default.
- W4310921857 hasConcept C62520636 @default.
- W4310921857 hasConcept C66322947 @default.
- W4310921857 hasConcept C89600930 @default.
- W4310921857 hasConcept C94625758 @default.
- W4310921857 hasConceptScore W4310921857C119857082 @default.
- W4310921857 hasConceptScore W4310921857C121332964 @default.
- W4310921857 hasConceptScore W4310921857C153180895 @default.
- W4310921857 hasConceptScore W4310921857C154945302 @default.
- W4310921857 hasConceptScore W4310921857C160633673 @default.
- W4310921857 hasConceptScore W4310921857C165801399 @default.
- W4310921857 hasConceptScore W4310921857C17744445 @default.
- W4310921857 hasConceptScore W4310921857C184337299 @default.
- W4310921857 hasConceptScore W4310921857C199360897 @default.
- W4310921857 hasConceptScore W4310921857C199539241 @default.
- W4310921857 hasConceptScore W4310921857C204321447 @default.
- W4310921857 hasConceptScore W4310921857C2776359362 @default.
- W4310921857 hasConceptScore W4310921857C2781238097 @default.
- W4310921857 hasConceptScore W4310921857C41008148 @default.
- W4310921857 hasConceptScore W4310921857C62520636 @default.
- W4310921857 hasConceptScore W4310921857C66322947 @default.
- W4310921857 hasConceptScore W4310921857C89600930 @default.
- W4310921857 hasConceptScore W4310921857C94625758 @default.
- W4310921857 hasLocation W43109218571 @default.
- W4310921857 hasLocation W43109218572 @default.
- W4310921857 hasOpenAccess W4310921857 @default.
- W4310921857 hasPrimaryLocation W43109218571 @default.
- W4310921857 hasRelatedWork W1982025852 @default.
- W4310921857 hasRelatedWork W2047970610 @default.
- W4310921857 hasRelatedWork W2053596378 @default.
- W4310921857 hasRelatedWork W2085033728 @default.
- W4310921857 hasRelatedWork W2106540031 @default.
- W4310921857 hasRelatedWork W2168523118 @default.
- W4310921857 hasRelatedWork W2171299904 @default.
- W4310921857 hasRelatedWork W2922442631 @default.
- W4310921857 hasRelatedWork W3112772842 @default.
- W4310921857 hasRelatedWork W4285411112 @default.
- W4310921857 isParatext "false" @default.
- W4310921857 isRetracted "false" @default.
- W4310921857 workType "article" @default.