Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386075498> ?p ?o ?g. }
Showing items 1 to 87 of
87
with 100 items per page.
- W4386075498 abstract "We propose a margin-based loss for tuning joint vision-language models so that their gradient-based explanations are consistent with region-level annotations provided by humans for relatively smaller grounding datasets. We refer to this objective as Attention Mask Consistency (AMC) and demonstrate that it produces superior visual grounding results than previous methods that rely on using vision-language models to score the outputs of object detectors. Particularly, a model trained with AMC on top of standard vision-language modeling objectives obtains a state-of-the-art accuracy of 86.49% in the Flickr30k visual grounding benchmark, an absolute improvement of 5.38% when compared to the best previous model trained under the same level of supervision. Our approach also performs exceedingly well on established benchmarks for referring expression comprehension where it obtains 80.34% accuracy in the easy test of RefCOCO+, and 64.55% in the difficult split. AMC is effective, easy to implement, and is general as it can be adopted by any vision-language model, and can use any type of region annotations." @default.
- W4386075498 created "2023-08-23" @default.
- W4386075498 creator A5027328044 @default.
- W4386075498 creator A5028863551 @default.
- W4386075498 creator A5029083062 @default.
- W4386075498 creator A5047193437 @default.
- W4386075498 date "2023-06-01" @default.
- W4386075498 modified "2023-09-27" @default.
- W4386075498 title "Improving Visual Grounding by Encouraging Consistent Gradient-Based Explanations" @default.
- W4386075498 cites W1773149199 @default.
- W4386075498 cites W2251512949 @default.
- W4386075498 cites W2277195237 @default.
- W4386075498 cites W2295107390 @default.
- W4386075498 cites W2745461083 @default.
- W4386075498 cites W2886641317 @default.
- W4386075498 cites W2964345792 @default.
- W4386075498 cites W2983256121 @default.
- W4386075498 cites W2989176720 @default.
- W4386075498 cites W3034727271 @default.
- W4386075498 cites W3100393531 @default.
- W4386075498 cites W3168796319 @default.
- W4386075498 cites W3172112830 @default.
- W4386075498 cites W3176641147 @default.
- W4386075498 cites W3178418424 @default.
- W4386075498 cites W3204474617 @default.
- W4386075498 cites W3213106580 @default.
- W4386075498 cites W4214650614 @default.
- W4386075498 cites W4230405732 @default.
- W4386075498 cites W4312956471 @default.
- W4386075498 doi "https://doi.org/10.1109/cvpr52729.2023.01837" @default.
- W4386075498 hasPublicationYear "2023" @default.
- W4386075498 type Work @default.
- W4386075498 citedByCount "0" @default.
- W4386075498 crossrefType "proceedings-article" @default.
- W4386075498 hasAuthorship W4386075498A5027328044 @default.
- W4386075498 hasAuthorship W4386075498A5028863551 @default.
- W4386075498 hasAuthorship W4386075498A5029083062 @default.
- W4386075498 hasAuthorship W4386075498A5047193437 @default.
- W4386075498 hasConcept C119857082 @default.
- W4386075498 hasConcept C121332964 @default.
- W4386075498 hasConcept C13280743 @default.
- W4386075498 hasConcept C137293760 @default.
- W4386075498 hasConcept C154945302 @default.
- W4386075498 hasConcept C168993435 @default.
- W4386075498 hasConcept C185798385 @default.
- W4386075498 hasConcept C199360897 @default.
- W4386075498 hasConcept C204321447 @default.
- W4386075498 hasConcept C205649164 @default.
- W4386075498 hasConcept C2776436953 @default.
- W4386075498 hasConcept C2781238097 @default.
- W4386075498 hasConcept C41008148 @default.
- W4386075498 hasConcept C511192102 @default.
- W4386075498 hasConcept C62520636 @default.
- W4386075498 hasConcept C774472 @default.
- W4386075498 hasConceptScore W4386075498C119857082 @default.
- W4386075498 hasConceptScore W4386075498C121332964 @default.
- W4386075498 hasConceptScore W4386075498C13280743 @default.
- W4386075498 hasConceptScore W4386075498C137293760 @default.
- W4386075498 hasConceptScore W4386075498C154945302 @default.
- W4386075498 hasConceptScore W4386075498C168993435 @default.
- W4386075498 hasConceptScore W4386075498C185798385 @default.
- W4386075498 hasConceptScore W4386075498C199360897 @default.
- W4386075498 hasConceptScore W4386075498C204321447 @default.
- W4386075498 hasConceptScore W4386075498C205649164 @default.
- W4386075498 hasConceptScore W4386075498C2776436953 @default.
- W4386075498 hasConceptScore W4386075498C2781238097 @default.
- W4386075498 hasConceptScore W4386075498C41008148 @default.
- W4386075498 hasConceptScore W4386075498C511192102 @default.
- W4386075498 hasConceptScore W4386075498C62520636 @default.
- W4386075498 hasConceptScore W4386075498C774472 @default.
- W4386075498 hasFunder F4320306076 @default.
- W4386075498 hasLocation W43860754981 @default.
- W4386075498 hasOpenAccess W4386075498 @default.
- W4386075498 hasPrimaryLocation W43860754981 @default.
- W4386075498 hasRelatedWork W1485630101 @default.
- W4386075498 hasRelatedWork W2359001871 @default.
- W4386075498 hasRelatedWork W2498017833 @default.
- W4386075498 hasRelatedWork W2961085424 @default.
- W4386075498 hasRelatedWork W2983785000 @default.
- W4386075498 hasRelatedWork W3116295307 @default.
- W4386075498 hasRelatedWork W3129868498 @default.
- W4386075498 hasRelatedWork W4281395811 @default.
- W4386075498 hasRelatedWork W4309086292 @default.
- W4386075498 hasRelatedWork W4386075498 @default.
- W4386075498 isParatext "false" @default.
- W4386075498 isRetracted "false" @default.
- W4386075498 workType "article" @default.