Matches in SemOpenAlex for { <https://semopenalex.org/work/W3133684245> ?p ?o ?g. }
- W3133684245 endingPage "1276" @default.
- W3133684245 startingPage "1266" @default.
- W3133684245 abstract "Scene graphs connect individual objects with visual relationships. They serve as a comprehensive scene representation for downstream multimodal tasks. However, by exploring recent progress in Scene Graph Generation (SGG), we find that the performance of recent works is highly limited by the pairwise relationship modeling by naive feature concatenation. Such pairwise features lack sufficient object interaction due to the mis-aligned object parts, resulting in non-discriminative pairwise features for visual relationship prediction. For example, naive concatenated pairwise feature usually make the model fail to discriminate between <monospace xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>riding</monospace> and <monospace xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>feeding</monospace> for object pair <monospace xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>person</monospace> and <monospace xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>horse</monospace> . To this end, we design a meta-architecture— learning-to-align — for dynamic object feature concatenation. We call our model: <bold xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>Align R-CNN</b> . Specifically, we introduce a novel attention-based multiple region alignment module that can be jointly optimized with SGG. Experiments on the large-scale SGG benchmark Visual Genome show that the proposed <bold xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>Align R-CNN</b> can replace the naive feature concatenation and thus boost all the existing SGG methods." @default.
- W3133684245 created "2021-03-15" @default.
- W3133684245 creator A5017130020 @default.
- W3133684245 creator A5042324027 @default.
- W3133684245 creator A5044120384 @default.
- W3133684245 creator A5077258552 @default.
- W3133684245 date "2022-01-01" @default.
- W3133684245 modified "2023-09-23" @default.
- W3133684245 title "Align R-CNN: A Pairwise Head Network for Visual Relationship Detection" @default.
- W3133684245 cites W1933349210 @default.
- W3133684245 cites W2077069816 @default.
- W3133684245 cites W2250378130 @default.
- W3133684245 cites W2277195237 @default.
- W3133684245 cites W2479423890 @default.
- W3133684245 cites W2481240925 @default.
- W3133684245 cites W2549139847 @default.
- W3133684245 cites W2565639579 @default.
- W3133684245 cites W2579549467 @default.
- W3133684245 cites W2591644541 @default.
- W3133684245 cites W2605736949 @default.
- W3133684245 cites W2607855566 @default.
- W3133684245 cites W2739107216 @default.
- W3133684245 cites W2740962769 @default.
- W3133684245 cites W2754191212 @default.
- W3133684245 cites W2795378892 @default.
- W3133684245 cites W2799029629 @default.
- W3133684245 cites W2806070179 @default.
- W3133684245 cites W2810482788 @default.
- W3133684245 cites W2883170015 @default.
- W3133684245 cites W2886970679 @default.
- W3133684245 cites W2887029921 @default.
- W3133684245 cites W2890531016 @default.
- W3133684245 cites W2913618459 @default.
- W3133684245 cites W2950096400 @default.
- W3133684245 cites W2955988340 @default.
- W3133684245 cites W2962764817 @default.
- W3133684245 cites W2962779575 @default.
- W3133684245 cites W2962785943 @default.
- W3133684245 cites W2962964995 @default.
- W3133684245 cites W2963091558 @default.
- W3133684245 cites W2963101956 @default.
- W3133684245 cites W2963514444 @default.
- W3133684245 cites W2963536419 @default.
- W3133684245 cites W2963649796 @default.
- W3133684245 cites W2963871344 @default.
- W3133684245 cites W2963902384 @default.
- W3133684245 cites W2963938081 @default.
- W3133684245 cites W2963980128 @default.
- W3133684245 cites W2964094751 @default.
- W3133684245 cites W2971974407 @default.
- W3133684245 cites W2976818183 @default.
- W3133684245 cites W2981967134 @default.
- W3133684245 cites W2987327987 @default.
- W3133684245 cites W2990356824 @default.
- W3133684245 cites W2992195701 @default.
- W3133684245 cites W3026441479 @default.
- W3133684245 cites W3035017890 @default.
- W3133684245 cites W3108864070 @default.
- W3133684245 cites W639708223 @default.
- W3133684245 doi "https://doi.org/10.1109/tmm.2021.3062543" @default.
- W3133684245 hasPublicationYear "2022" @default.
- W3133684245 type Work @default.
- W3133684245 sameAs 3133684245 @default.
- W3133684245 citedByCount "4" @default.
- W3133684245 countsByYear W31336842452022 @default.
- W3133684245 countsByYear W31336842452023 @default.
- W3133684245 crossrefType "journal-article" @default.
- W3133684245 hasAuthorship W3133684245A5017130020 @default.
- W3133684245 hasAuthorship W3133684245A5042324027 @default.
- W3133684245 hasAuthorship W3133684245A5044120384 @default.
- W3133684245 hasAuthorship W3133684245A5077258552 @default.
- W3133684245 hasConcept C114614502 @default.
- W3133684245 hasConcept C138885662 @default.
- W3133684245 hasConcept C153180895 @default.
- W3133684245 hasConcept C154945302 @default.
- W3133684245 hasConcept C184898388 @default.
- W3133684245 hasConcept C204321447 @default.
- W3133684245 hasConcept C2776401178 @default.
- W3133684245 hasConcept C33923547 @default.
- W3133684245 hasConcept C36464697 @default.
- W3133684245 hasConcept C41008148 @default.
- W3133684245 hasConcept C41895202 @default.
- W3133684245 hasConcept C87619178 @default.
- W3133684245 hasConcept C97931131 @default.
- W3133684245 hasConceptScore W3133684245C114614502 @default.
- W3133684245 hasConceptScore W3133684245C138885662 @default.
- W3133684245 hasConceptScore W3133684245C153180895 @default.
- W3133684245 hasConceptScore W3133684245C154945302 @default.
- W3133684245 hasConceptScore W3133684245C184898388 @default.
- W3133684245 hasConceptScore W3133684245C204321447 @default.
- W3133684245 hasConceptScore W3133684245C2776401178 @default.
- W3133684245 hasConceptScore W3133684245C33923547 @default.
- W3133684245 hasConceptScore W3133684245C36464697 @default.
- W3133684245 hasConceptScore W3133684245C41008148 @default.
- W3133684245 hasConceptScore W3133684245C41895202 @default.
- W3133684245 hasConceptScore W3133684245C87619178 @default.
- W3133684245 hasConceptScore W3133684245C97931131 @default.
- W3133684245 hasLocation W31336842451 @default.