Matches in SemOpenAlex for { <https://semopenalex.org/work/W4312920106> ?p ?o ?g. }
- W4312920106 abstract "We introduce Retrieval Augmented Classification (RAC), a generic approach to augmenting standard image classification pipelines with an explicit retrieval module. RAC consists of a standard base image encoder fused with a parallel retrieval branch that queries a non-parametric external memory of pre-encoded images and associated text snippets. We apply RAC to the problem of long-tail classification and demonstrate a significant improvement over previous state-of-the-art on Places365-LT and iNaturalist-2018 (14.5% and 6.7% respectively), despite using only the training datasets themselves as the external information source. We demonstrate that RAC's retrieval module, without prompting, learns a high level of accuracy on tail classes. This, in turn, frees the base encoder to focus on common classes, and improve its performance thereon. RAC represents an alternative approach to utilizing large, pretrained models without requiring fine-tuning, as well as a first step towards more effectively making use of external memory within common computer vision architectures." @default.
- W4312920106 created "2023-01-05" @default.
- W4312920106 creator A5006294869 @default.
- W4312920106 creator A5014702970 @default.
- W4312920106 creator A5026101344 @default.
- W4312920106 creator A5028024287 @default.
- W4312920106 creator A5029106652 @default.
- W4312920106 creator A5033143541 @default.
- W4312920106 creator A5039055995 @default.
- W4312920106 creator A5078883821 @default.
- W4312920106 creator A5091407554 @default.
- W4312920106 date "2022-06-01" @default.
- W4312920106 modified "2023-10-10" @default.
- W4312920106 title "Retrieval Augmented Classification for Long-Tail Visual Recognition" @default.
- W4312920106 cites W1582392998 @default.
- W4312920106 cites W2023954349 @default.
- W4312920106 cites W2108598243 @default.
- W4312920106 cites W2118978333 @default.
- W4312920106 cites W2132791018 @default.
- W4312920106 cites W2148143831 @default.
- W4312920106 cites W2183341477 @default.
- W4312920106 cites W2250539671 @default.
- W4312920106 cites W2440599146 @default.
- W4312920106 cites W2732026016 @default.
- W4312920106 cites W2781292787 @default.
- W4312920106 cites W2797977484 @default.
- W4312920106 cites W2951434086 @default.
- W4312920106 cites W2962843773 @default.
- W4312920106 cites W2962933664 @default.
- W4312920106 cites W2963351448 @default.
- W4312920106 cites W2963469388 @default.
- W4312920106 cites W2963691377 @default.
- W4312920106 cites W2963703197 @default.
- W4312920106 cites W2987228832 @default.
- W4312920106 cites W3034369739 @default.
- W4312920106 cites W3034601242 @default.
- W4312920106 cites W3096121526 @default.
- W4312920106 cites W3096688134 @default.
- W4312920106 cites W3097217077 @default.
- W4312920106 cites W3099700870 @default.
- W4312920106 cites W3109241881 @default.
- W4312920106 cites W3166596953 @default.
- W4312920106 cites W3169726359 @default.
- W4312920106 cites W3177200443 @default.
- W4312920106 cites W3177357839 @default.
- W4312920106 cites W3179096061 @default.
- W4312920106 cites W4206118214 @default.
- W4312920106 cites W4214718285 @default.
- W4312920106 doi "https://doi.org/10.1109/cvpr52688.2022.00683" @default.
- W4312920106 hasPublicationYear "2022" @default.
- W4312920106 type Work @default.
- W4312920106 citedByCount "9" @default.
- W4312920106 countsByYear W43129201062022 @default.
- W4312920106 countsByYear W43129201062023 @default.
- W4312920106 crossrefType "proceedings-article" @default.
- W4312920106 hasAuthorship W4312920106A5006294869 @default.
- W4312920106 hasAuthorship W4312920106A5014702970 @default.
- W4312920106 hasAuthorship W4312920106A5026101344 @default.
- W4312920106 hasAuthorship W4312920106A5028024287 @default.
- W4312920106 hasAuthorship W4312920106A5029106652 @default.
- W4312920106 hasAuthorship W4312920106A5033143541 @default.
- W4312920106 hasAuthorship W4312920106A5039055995 @default.
- W4312920106 hasAuthorship W4312920106A5078883821 @default.
- W4312920106 hasAuthorship W4312920106A5091407554 @default.
- W4312920106 hasBestOaLocation W43129201062 @default.
- W4312920106 hasConcept C105795698 @default.
- W4312920106 hasConcept C111919701 @default.
- W4312920106 hasConcept C115961682 @default.
- W4312920106 hasConcept C117251300 @default.
- W4312920106 hasConcept C118505674 @default.
- W4312920106 hasConcept C120665830 @default.
- W4312920106 hasConcept C121332964 @default.
- W4312920106 hasConcept C134306372 @default.
- W4312920106 hasConcept C153180895 @default.
- W4312920106 hasConcept C154945302 @default.
- W4312920106 hasConcept C1667742 @default.
- W4312920106 hasConcept C192209626 @default.
- W4312920106 hasConcept C23123220 @default.
- W4312920106 hasConcept C33923547 @default.
- W4312920106 hasConcept C41008148 @default.
- W4312920106 hasConcept C42058472 @default.
- W4312920106 hasConceptScore W4312920106C105795698 @default.
- W4312920106 hasConceptScore W4312920106C111919701 @default.
- W4312920106 hasConceptScore W4312920106C115961682 @default.
- W4312920106 hasConceptScore W4312920106C117251300 @default.
- W4312920106 hasConceptScore W4312920106C118505674 @default.
- W4312920106 hasConceptScore W4312920106C120665830 @default.
- W4312920106 hasConceptScore W4312920106C121332964 @default.
- W4312920106 hasConceptScore W4312920106C134306372 @default.
- W4312920106 hasConceptScore W4312920106C153180895 @default.
- W4312920106 hasConceptScore W4312920106C154945302 @default.
- W4312920106 hasConceptScore W4312920106C1667742 @default.
- W4312920106 hasConceptScore W4312920106C192209626 @default.
- W4312920106 hasConceptScore W4312920106C23123220 @default.
- W4312920106 hasConceptScore W4312920106C33923547 @default.
- W4312920106 hasConceptScore W4312920106C41008148 @default.
- W4312920106 hasConceptScore W4312920106C42058472 @default.
- W4312920106 hasLocation W43129201061 @default.
- W4312920106 hasLocation W43129201062 @default.
- W4312920106 hasOpenAccess W4312920106 @default.