Matches in SemOpenAlex for { <https://semopenalex.org/work/W2550980560> ?p ?o ?g. }
- W2550980560 abstract "Lipreading is the task of decoding text from the movement of a speaker'smouth. Traditional approaches separated the problem into two stages: designingor learning visual features, and prediction. More recent deep lipreadingapproaches are end-to-end trainable (Wand et al., 2016; Chung & Zisserman,2016a). All existing works, however, perform only word classification, notsentence-level sequence prediction. Studies have shown that human lipreadingperformance increases for longer words (Easton & Basala, 1982), indicating theimportance of features capturing temporal context in an ambiguous communicationchannel. Motivated by this observation, we present LipNet, a model that maps avariable-length sequence of video frames to text, making use of spatiotemporalconvolutions, an LSTM recurrent network, and the connectionist temporalclassification loss, trained entirely end-to-end. To the best of our knowledge,LipNet is the first lipreading model to operate at sentence-level, using asingle end-to-end speaker-independent deep model to simultaneously learnspatiotemporal visual features and a sequence model. On the GRID corpus, LipNetachieves 93.4% accuracy, outperforming experienced human lipreaders and theprevious 79.6% state-of-the-art accuracy." @default.
- W2550980560 created "2016-11-30" @default.
- W2550980560 creator A5012498862 @default.
- W2550980560 creator A5043118390 @default.
- W2550980560 creator A5056879203 @default.
- W2550980560 creator A5082304130 @default.
- W2550980560 date "2016-11-05" @default.
- W2550980560 modified "2023-10-01" @default.
- W2550980560 title "LipNet: Sentence-level Lipreading." @default.
- W2550980560 cites W1211870555 @default.
- W2550980560 cites W1558887665 @default.
- W2550980560 cites W1849277567 @default.
- W2550980560 cites W2000399372 @default.
- W2550980560 cites W2015143272 @default.
- W2550980560 cites W2015394094 @default.
- W2550980560 cites W2026028405 @default.
- W2550980560 cites W2043181877 @default.
- W2550980560 cites W2058961190 @default.
- W2550980560 cites W2060510034 @default.
- W2550980560 cites W2064675550 @default.
- W2550980560 cites W2071096833 @default.
- W2550980560 cites W2102113734 @default.
- W2550980560 cites W2107435954 @default.
- W2550980560 cites W2111006958 @default.
- W2550980560 cites W2113814270 @default.
- W2550980560 cites W2115252128 @default.
- W2550980560 cites W2124397839 @default.
- W2550980560 cites W2127141656 @default.
- W2550980560 cites W2136155248 @default.
- W2550980560 cites W2142075667 @default.
- W2550980560 cites W2147768505 @default.
- W2550980560 cites W2160815625 @default.
- W2550980560 cites W2171209807 @default.
- W2550980560 cites W2184188583 @default.
- W2550980560 cites W2193413348 @default.
- W2550980560 cites W2243738093 @default.
- W2550980560 cites W2267805933 @default.
- W2550980560 cites W2293858598 @default.
- W2550980560 cites W2308045930 @default.
- W2550980560 cites W2398406965 @default.
- W2550980560 cites W2516412152 @default.
- W2550980560 cites W2556171197 @default.
- W2550980560 cites W2594690981 @default.
- W2550980560 cites W2604379605 @default.
- W2550980560 cites W2612434330 @default.
- W2550980560 cites W2962851944 @default.
- W2550980560 cites W2963382180 @default.
- W2550980560 cites W3099202502 @default.
- W2550980560 hasPublicationYear "2016" @default.
- W2550980560 type Work @default.
- W2550980560 sameAs 2550980560 @default.
- W2550980560 citedByCount "30" @default.
- W2550980560 countsByYear W25509805602017 @default.
- W2550980560 countsByYear W25509805602018 @default.
- W2550980560 countsByYear W25509805602019 @default.
- W2550980560 countsByYear W25509805602020 @default.
- W2550980560 countsByYear W25509805602021 @default.
- W2550980560 crossrefType "posted-content" @default.
- W2550980560 hasAuthorship W2550980560A5012498862 @default.
- W2550980560 hasAuthorship W2550980560A5043118390 @default.
- W2550980560 hasAuthorship W2550980560A5056879203 @default.
- W2550980560 hasAuthorship W2550980560A5082304130 @default.
- W2550980560 hasConcept C138885662 @default.
- W2550980560 hasConcept C151730666 @default.
- W2550980560 hasConcept C154945302 @default.
- W2550980560 hasConcept C162324750 @default.
- W2550980560 hasConcept C187736073 @default.
- W2550980560 hasConcept C204321447 @default.
- W2550980560 hasConcept C2777530160 @default.
- W2550980560 hasConcept C2778112365 @default.
- W2550980560 hasConcept C2779343474 @default.
- W2550980560 hasConcept C2780451532 @default.
- W2550980560 hasConcept C28490314 @default.
- W2550980560 hasConcept C40506919 @default.
- W2550980560 hasConcept C41008148 @default.
- W2550980560 hasConcept C41895202 @default.
- W2550980560 hasConcept C50644808 @default.
- W2550980560 hasConcept C54355233 @default.
- W2550980560 hasConcept C57273362 @default.
- W2550980560 hasConcept C76155785 @default.
- W2550980560 hasConcept C8521452 @default.
- W2550980560 hasConcept C86803240 @default.
- W2550980560 hasConcept C90805587 @default.
- W2550980560 hasConceptScore W2550980560C138885662 @default.
- W2550980560 hasConceptScore W2550980560C151730666 @default.
- W2550980560 hasConceptScore W2550980560C154945302 @default.
- W2550980560 hasConceptScore W2550980560C162324750 @default.
- W2550980560 hasConceptScore W2550980560C187736073 @default.
- W2550980560 hasConceptScore W2550980560C204321447 @default.
- W2550980560 hasConceptScore W2550980560C2777530160 @default.
- W2550980560 hasConceptScore W2550980560C2778112365 @default.
- W2550980560 hasConceptScore W2550980560C2779343474 @default.
- W2550980560 hasConceptScore W2550980560C2780451532 @default.
- W2550980560 hasConceptScore W2550980560C28490314 @default.
- W2550980560 hasConceptScore W2550980560C40506919 @default.
- W2550980560 hasConceptScore W2550980560C41008148 @default.
- W2550980560 hasConceptScore W2550980560C41895202 @default.
- W2550980560 hasConceptScore W2550980560C50644808 @default.
- W2550980560 hasConceptScore W2550980560C54355233 @default.
- W2550980560 hasConceptScore W2550980560C57273362 @default.