Matches in SemOpenAlex for { <https://semopenalex.org/work/W4367692162> ?p ?o ?g. }
Showing items 1 to 59 of
59
with 100 items per page.
- W4367692162 abstract "Automated image captioning has the potential to be a useful tool for people with vision impairments. Images taken by this user group are often noisy, which leads to incorrect and even unsafe model predictions. In this paper, we propose a quality-agnostic framework to improve the performance and robustness of image captioning models for visually impaired people. We address this problem from three angles: data, model, and evaluation. First, we show how data augmentation techniques for generating synthetic noise can address data sparsity in this domain. Second, we enhance the robustness of the model by expanding a state-of-the-art model to a dual network architecture, using the augmented data and leveraging different consistency losses. Our results demonstrate increased performance, e.g. an absolute improvement of 2.15 on CIDEr, compared to state-of-the-art image captioning networks, as well as increased robustness to noise with up to 3 points improvement on CIDEr in more noisy settings. Finally, we evaluate the prediction reliability using confidence calibration on images with different difficulty/noise levels, showing that our models perform more reliably in safety-critical situations. The improved model is part of an assisted living application, which we develop in partnership with the Royal National Institute of Blind People." @default.
- W4367692162 created "2023-05-03" @default.
- W4367692162 creator A5029718321 @default.
- W4367692162 creator A5035064408 @default.
- W4367692162 creator A5074032387 @default.
- W4367692162 creator A5076593865 @default.
- W4367692162 date "2023-04-28" @default.
- W4367692162 modified "2023-10-16" @default.
- W4367692162 title "Quality-agnostic Image Captioning to Safely Assist People with Vision Impairment" @default.
- W4367692162 doi "https://doi.org/10.48550/arxiv.2304.14623" @default.
- W4367692162 hasPublicationYear "2023" @default.
- W4367692162 type Work @default.
- W4367692162 citedByCount "0" @default.
- W4367692162 crossrefType "posted-content" @default.
- W4367692162 hasAuthorship W4367692162A5029718321 @default.
- W4367692162 hasAuthorship W4367692162A5035064408 @default.
- W4367692162 hasAuthorship W4367692162A5074032387 @default.
- W4367692162 hasAuthorship W4367692162A5076593865 @default.
- W4367692162 hasBestOaLocation W43676921621 @default.
- W4367692162 hasConcept C104317684 @default.
- W4367692162 hasConcept C115961682 @default.
- W4367692162 hasConcept C119857082 @default.
- W4367692162 hasConcept C154945302 @default.
- W4367692162 hasConcept C157657479 @default.
- W4367692162 hasConcept C185592680 @default.
- W4367692162 hasConcept C31972630 @default.
- W4367692162 hasConcept C41008148 @default.
- W4367692162 hasConcept C55020928 @default.
- W4367692162 hasConcept C55493867 @default.
- W4367692162 hasConcept C63479239 @default.
- W4367692162 hasConcept C99498987 @default.
- W4367692162 hasConceptScore W4367692162C104317684 @default.
- W4367692162 hasConceptScore W4367692162C115961682 @default.
- W4367692162 hasConceptScore W4367692162C119857082 @default.
- W4367692162 hasConceptScore W4367692162C154945302 @default.
- W4367692162 hasConceptScore W4367692162C157657479 @default.
- W4367692162 hasConceptScore W4367692162C185592680 @default.
- W4367692162 hasConceptScore W4367692162C31972630 @default.
- W4367692162 hasConceptScore W4367692162C41008148 @default.
- W4367692162 hasConceptScore W4367692162C55020928 @default.
- W4367692162 hasConceptScore W4367692162C55493867 @default.
- W4367692162 hasConceptScore W4367692162C63479239 @default.
- W4367692162 hasConceptScore W4367692162C99498987 @default.
- W4367692162 hasLocation W43676921621 @default.
- W4367692162 hasOpenAccess W4367692162 @default.
- W4367692162 hasPrimaryLocation W43676921621 @default.
- W4367692162 hasRelatedWork W2035976912 @default.
- W4367692162 hasRelatedWork W2036807459 @default.
- W4367692162 hasRelatedWork W2109974539 @default.
- W4367692162 hasRelatedWork W2125927971 @default.
- W4367692162 hasRelatedWork W2541791370 @default.
- W4367692162 hasRelatedWork W2574052219 @default.
- W4367692162 hasRelatedWork W2738084969 @default.
- W4367692162 hasRelatedWork W2795359650 @default.
- W4367692162 hasRelatedWork W2923366293 @default.
- W4367692162 hasRelatedWork W3008515501 @default.
- W4367692162 isParatext "false" @default.
- W4367692162 isRetracted "false" @default.
- W4367692162 workType "article" @default.