Matches in SemOpenAlex for { <https://semopenalex.org/work/W2963714798> ?p ?o ?g. }
Showing items 1 to 59 of
59
with 100 items per page.
- W2963714798 abstract "The interaction between language and visual information has been emphasizedin visual question answering (VQA) with the help of attention mechanism.However, the relationship between words in question has been underestimated,which makes it hard to answer questions that involve the relationship betweenmultiple entities, such as comparison and counting. In this paper, we developthe graph reasoning networks to tackle this problem. Two kinds of graphs areinvestigated, namely inter-graph and intra-graph. The inter-graph transfersfeatures of the detected objects to their related query words, enabling theoutput nodes to have both semantic and factual information. The intra-graphexchanges information between these output nodes from inter-graph to amplifyimplicit yet important relationship between objects. These two kinds of graphscooperate with each other, and thus our resulting model can reason therelationship and dependence between objects, which leads to realization ofmulti-step reasoning. Experimental results on the GQA v1.1 dataset demonstratethe reasoning ability of our method to handle compositional questions aboutreal-world images. We achieve state-of-the-art performance, boosting accuracyto 57.04%. On the VQA 2.0 dataset, we also receive a promising improvement onoverall accuracy, especially on counting problem." @default.
- W2963714798 created "2019-07-30" @default.
- W2963714798 creator A5001529504 @default.
- W2963714798 creator A5001819736 @default.
- W2963714798 creator A5032623398 @default.
- W2963714798 date "2019-07-23" @default.
- W2963714798 modified "2023-09-27" @default.
- W2963714798 title "Graph Reasoning Networks for Visual Question Answering." @default.
- W2963714798 cites W2745461083 @default.
- W2963714798 hasPublicationYear "2019" @default.
- W2963714798 type Work @default.
- W2963714798 sameAs 2963714798 @default.
- W2963714798 citedByCount "3" @default.
- W2963714798 countsByYear W29637147982018 @default.
- W2963714798 countsByYear W29637147982019 @default.
- W2963714798 countsByYear W29637147982021 @default.
- W2963714798 crossrefType "posted-content" @default.
- W2963714798 hasAuthorship W2963714798A5001529504 @default.
- W2963714798 hasAuthorship W2963714798A5001819736 @default.
- W2963714798 hasAuthorship W2963714798A5032623398 @default.
- W2963714798 hasConcept C132525143 @default.
- W2963714798 hasConcept C154945302 @default.
- W2963714798 hasConcept C41008148 @default.
- W2963714798 hasConcept C44291984 @default.
- W2963714798 hasConcept C46686674 @default.
- W2963714798 hasConcept C80444323 @default.
- W2963714798 hasConceptScore W2963714798C132525143 @default.
- W2963714798 hasConceptScore W2963714798C154945302 @default.
- W2963714798 hasConceptScore W2963714798C41008148 @default.
- W2963714798 hasConceptScore W2963714798C44291984 @default.
- W2963714798 hasConceptScore W2963714798C46686674 @default.
- W2963714798 hasConceptScore W2963714798C80444323 @default.
- W2963714798 hasLocation W29637147981 @default.
- W2963714798 hasOpenAccess W2963714798 @default.
- W2963714798 hasPrimaryLocation W29637147981 @default.
- W2963714798 hasRelatedWork W2243122402 @default.
- W2963714798 hasRelatedWork W2745461083 @default.
- W2963714798 hasRelatedWork W2807505183 @default.
- W2963714798 hasRelatedWork W2869186267 @default.
- W2963714798 hasRelatedWork W2963051286 @default.
- W2963714798 hasRelatedWork W2963954913 @default.
- W2963714798 hasRelatedWork W3005672611 @default.
- W2963714798 hasRelatedWork W3035075850 @default.
- W2963714798 hasRelatedWork W3036683340 @default.
- W2963714798 hasRelatedWork W3092782620 @default.
- W2963714798 hasRelatedWork W3093153624 @default.
- W2963714798 hasRelatedWork W3096975985 @default.
- W2963714798 hasRelatedWork W3125115207 @default.
- W2963714798 hasRelatedWork W3175147232 @default.
- W2963714798 hasRelatedWork W3175445769 @default.
- W2963714798 hasRelatedWork W3193810005 @default.
- W2963714798 hasRelatedWork W2971455676 @default.
- W2963714798 hasRelatedWork W3081814550 @default.
- W2963714798 hasRelatedWork W3101023622 @default.
- W2963714798 hasRelatedWork W3107622140 @default.
- W2963714798 isParatext "false" @default.
- W2963714798 isRetracted "false" @default.
- W2963714798 magId "2963714798" @default.
- W2963714798 workType "article" @default.