Matches in SemOpenAlex for { <https://semopenalex.org/work/W3110157234> ?p ?o ?g. }
- W3110157234 endingPage "644" @default.
- W3110157234 startingPage "629" @default.
- W3110157234 abstract "Despite considerable progress, state of the art image captioning models produce generic captions, leaving out important image details. Furthermore, these systems may even misrepresent the image in order to produce a simpler caption consisting of common concepts. In this paper, we first analyze both modern captioning systems and evaluation metrics through empirical experiments to quantify these phenomena. We find that modern captioning systems return higher likelihoods for incorrect distractor sentences compared to ground truth captions, and that evaluation metrics like SPICE can be ‘topped’ using simple captioning systems relying on object detectors. Inspired by these observations, we design a new metric (SPICE-U) by introducing a notion of uniqueness over the concepts generated in a caption. We show that SPICE-U is better correlated with human judgements compared to SPICE, and effectively captures notions of diversity and descriptiveness. Finally, we also demonstrate a general technique to improve any existing captioning model – by using mutual information as a re-ranking objective during decoding. Empirically, this results in more unique and informative captions, and improves three different state-of-the-art models on SPICE-U as well as average score over existing metrics (Code is available at https://github.com/princetonvisualai/SPICE-U )." @default.
- W3110157234 created "2020-12-07" @default.
- W3110157234 creator A5022811687 @default.
- W3110157234 creator A5025205227 @default.
- W3110157234 creator A5026985811 @default.
- W3110157234 creator A5038268246 @default.
- W3110157234 date "2020-01-01" @default.
- W3110157234 modified "2023-09-25" @default.
- W3110157234 title "Towards Unique and Informative Captioning of Images" @default.
- W3110157234 cites W1861492603 @default.
- W3110157234 cites W1895577753 @default.
- W3110157234 cites W1905882502 @default.
- W3110157234 cites W1956340063 @default.
- W3110157234 cites W1969616664 @default.
- W3110157234 cites W1999575133 @default.
- W3110157234 cites W2077069816 @default.
- W3110157234 cites W2101105183 @default.
- W3110157234 cites W2133512280 @default.
- W3110157234 cites W2277195237 @default.
- W3110157234 cites W2302086703 @default.
- W3110157234 cites W2463955103 @default.
- W3110157234 cites W2506483933 @default.
- W3110157234 cites W2529784951 @default.
- W3110157234 cites W2549599535 @default.
- W3110157234 cites W2574790321 @default.
- W3110157234 cites W2575842049 @default.
- W3110157234 cites W2604178507 @default.
- W3110157234 cites W2618264341 @default.
- W3110157234 cites W2738881192 @default.
- W3110157234 cites W2745461083 @default.
- W3110157234 cites W2788277448 @default.
- W3110157234 cites W2890781596 @default.
- W3110157234 cites W2893724244 @default.
- W3110157234 cites W2949376505 @default.
- W3110157234 cites W2954841306 @default.
- W3110157234 cites W2962735233 @default.
- W3110157234 cites W2963109634 @default.
- W3110157234 cites W2963138277 @default.
- W3110157234 cites W2963170456 @default.
- W3110157234 cites W2963206148 @default.
- W3110157234 cites W2963448089 @default.
- W3110157234 cites W2963551569 @default.
- W3110157234 cites W2963758027 @default.
- W3110157234 cites W2963966654 @default.
- W3110157234 cites W2964024144 @default.
- W3110157234 cites W2964042428 @default.
- W3110157234 cites W2965289598 @default.
- W3110157234 cites W2967223102 @default.
- W3110157234 cites W2986670728 @default.
- W3110157234 cites W2989489923 @default.
- W3110157234 cites W3149335959 @default.
- W3110157234 cites W639708223 @default.
- W3110157234 doi "https://doi.org/10.1007/978-3-030-58571-6_37" @default.
- W3110157234 hasPublicationYear "2020" @default.
- W3110157234 type Work @default.
- W3110157234 sameAs 3110157234 @default.
- W3110157234 citedByCount "16" @default.
- W3110157234 countsByYear W31101572342021 @default.
- W3110157234 countsByYear W31101572342022 @default.
- W3110157234 countsByYear W31101572342023 @default.
- W3110157234 crossrefType "book-chapter" @default.
- W3110157234 hasAuthorship W3110157234A5022811687 @default.
- W3110157234 hasAuthorship W3110157234A5025205227 @default.
- W3110157234 hasAuthorship W3110157234A5026985811 @default.
- W3110157234 hasAuthorship W3110157234A5038268246 @default.
- W3110157234 hasBestOaLocation W31101572342 @default.
- W3110157234 hasConcept C11413529 @default.
- W3110157234 hasConcept C115961682 @default.
- W3110157234 hasConcept C119599485 @default.
- W3110157234 hasConcept C127413603 @default.
- W3110157234 hasConcept C154945302 @default.
- W3110157234 hasConcept C157657479 @default.
- W3110157234 hasConcept C162324750 @default.
- W3110157234 hasConcept C176217482 @default.
- W3110157234 hasConcept C177264268 @default.
- W3110157234 hasConcept C189430467 @default.
- W3110157234 hasConcept C199360897 @default.
- W3110157234 hasConcept C204321447 @default.
- W3110157234 hasConcept C21547014 @default.
- W3110157234 hasConcept C23123220 @default.
- W3110157234 hasConcept C2776760102 @default.
- W3110157234 hasConcept C2780077345 @default.
- W3110157234 hasConcept C2781238097 @default.
- W3110157234 hasConcept C31170391 @default.
- W3110157234 hasConcept C34447519 @default.
- W3110157234 hasConcept C41008148 @default.
- W3110157234 hasConcept C57273362 @default.
- W3110157234 hasConceptScore W3110157234C11413529 @default.
- W3110157234 hasConceptScore W3110157234C115961682 @default.
- W3110157234 hasConceptScore W3110157234C119599485 @default.
- W3110157234 hasConceptScore W3110157234C127413603 @default.
- W3110157234 hasConceptScore W3110157234C154945302 @default.
- W3110157234 hasConceptScore W3110157234C157657479 @default.
- W3110157234 hasConceptScore W3110157234C162324750 @default.
- W3110157234 hasConceptScore W3110157234C176217482 @default.
- W3110157234 hasConceptScore W3110157234C177264268 @default.
- W3110157234 hasConceptScore W3110157234C189430467 @default.
- W3110157234 hasConceptScore W3110157234C199360897 @default.