FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4323306785> ?p ?o ?g. }

• W4323306785 endingPage "10711" @default.
• W4323306785 startingPage "10697" @default.
• W4323306785 abstract "Aiming the problem that the spatial-temporal hierarchical continuous sign language recognition (CSLR) model with video as input is computationally intensive, thus limiting the real-time application, this paper proposes a temporal super-resolution network (TSRNet) to reduce the model computation while keeping the loss of accuracy to a minimum, achieving the best compromise between the real-time performance and accuracy. The TSRNet-based CSLR constructed in this paper consists of three main parts: frame-level feature extraction, temporal feature extraction and the proposed TSRNet, where the TSRNet is located between them, and consists of two branches: detail and coarse descriptors. The extracted frame-level features are first sparse, after which they are passed through the two branches designed for feature reconstruction; the fused dense sequence is subjected to temporal feature extraction. In order to better recover the semantic-level information, this paper also proposes a self-generating adversarial network training method, which treats the TSRNet as the generator and the frame-level and temporal processing parts as the discriminator. In addition, to unify the criteria for judging the loss of model accuracy under different benchmarks, this paper proposes word error rate deviation (WERD), where the error rate between estimated WER and reference WER obtained by reconstructed frame-level feature sequence and complete original frame-level feature sequence, respectively. Experiments on two large-scale sign language datasets demonstrate the effectiveness of the model. The method proposed in this paper is not only applicable to CSLR, but is general to spatial-temporal hierarchical models where the input is video data. Code is available at https://github.com/woshisad159/CSLR.git ." @default.
• W4323306785 created "2023-03-07" @default.
• W4323306785 creator A5004219544 @default.
• W4323306785 creator A5041131074 @default.
• W4323306785 creator A5046553390 @default.
• W4323306785 creator A5048263320 @default.
• W4323306785 date "2023-03-06" @default.
• W4323306785 modified "2023-10-16" @default.
• W4323306785 title "Continuous Sign Language Recognition Via Temporal Super-Resolution Network" @default.
• W4323306785 cites W1885185971 @default.
• W4323306785 cites W2127141656 @default.
• W4323306785 cites W2188882108 @default.
• W4323306785 cites W2194775991 @default.
• W4323306785 cites W2507009361 @default.
• W4323306785 cites W2587277634 @default.
• W4323306785 cites W2755802490 @default.
• W4323306785 cites W2891726870 @default.
• W4323306785 cites W2895638065 @default.
• W4323306785 cites W2908497602 @default.
• W4323306785 cites W2913845868 @default.
• W4323306785 cites W2941870244 @default.
• W4323306785 cites W2963103155 @default.
• W4323306785 cites W2963524571 @default.
• W4323306785 cites W2964253156 @default.
• W4323306785 cites W2971915722 @default.
• W4323306785 cites W2997931247 @default.
• W4323306785 cites W3015734078 @default.
• W4323306785 cites W3033228789 @default.
• W4323306785 cites W3044723170 @default.
• W4323306785 cites W3046952127 @default.
• W4323306785 cites W3081930016 @default.
• W4323306785 cites W3082666011 @default.
• W4323306785 cites W3097462280 @default.
• W4323306785 cites W3113106473 @default.
• W4323306785 cites W3128176988 @default.
• W4323306785 cites W3139022918 @default.
• W4323306785 cites W3143289910 @default.
• W4323306785 cites W3147467731 @default.
• W4323306785 cites W3157193418 @default.
• W4323306785 cites W3160109012 @default.
• W4323306785 cites W3162420734 @default.
• W4323306785 cites W3184828860 @default.
• W4323306785 cites W3198876595 @default.
• W4323306785 cites W3206043672 @default.
• W4323306785 cites W3208937263 @default.
• W4323306785 cites W4200062228 @default.
• W4323306785 cites W4210359132 @default.
• W4323306785 cites W4213127165 @default.
• W4323306785 cites W4220752354 @default.
• W4323306785 cites W4280556793 @default.
• W4323306785 cites W4312443924 @default.
• W4323306785 doi "https://doi.org/10.1007/s13369-023-07718-8" @default.
• W4323306785 hasPublicationYear "2023" @default.
• W4323306785 type Work @default.
• W4323306785 citedByCount "0" @default.
• W4323306785 crossrefType "journal-article" @default.
• W4323306785 hasAuthorship W4323306785A5004219544 @default.
• W4323306785 hasAuthorship W4323306785A5041131074 @default.
• W4323306785 hasAuthorship W4323306785A5046553390 @default.
• W4323306785 hasAuthorship W4323306785A5048263320 @default.
• W4323306785 hasBestOaLocation W43233067852 @default.
• W4323306785 hasConcept C119666444 @default.
• W4323306785 hasConcept C121332964 @default.
• W4323306785 hasConcept C126042441 @default.
• W4323306785 hasConcept C138885662 @default.
• W4323306785 hasConcept C153180895 @default.
• W4323306785 hasConcept C154945302 @default.
• W4323306785 hasConcept C163258240 @default.
• W4323306785 hasConcept C2776401178 @default.
• W4323306785 hasConcept C2779803651 @default.
• W4323306785 hasConcept C2780992000 @default.
• W4323306785 hasConcept C3261483 @default.
• W4323306785 hasConcept C41008148 @default.
• W4323306785 hasConcept C41895202 @default.
• W4323306785 hasConcept C522192633 @default.
• W4323306785 hasConcept C52622490 @default.
• W4323306785 hasConcept C62520636 @default.
• W4323306785 hasConcept C76155785 @default.
• W4323306785 hasConcept C94915269 @default.
• W4323306785 hasConceptScore W4323306785C119666444 @default.
• W4323306785 hasConceptScore W4323306785C121332964 @default.
• W4323306785 hasConceptScore W4323306785C126042441 @default.
• W4323306785 hasConceptScore W4323306785C138885662 @default.
• W4323306785 hasConceptScore W4323306785C153180895 @default.
• W4323306785 hasConceptScore W4323306785C154945302 @default.
• W4323306785 hasConceptScore W4323306785C163258240 @default.
• W4323306785 hasConceptScore W4323306785C2776401178 @default.
• W4323306785 hasConceptScore W4323306785C2779803651 @default.
• W4323306785 hasConceptScore W4323306785C2780992000 @default.
• W4323306785 hasConceptScore W4323306785C3261483 @default.
• W4323306785 hasConceptScore W4323306785C41008148 @default.
• W4323306785 hasConceptScore W4323306785C41895202 @default.
• W4323306785 hasConceptScore W4323306785C522192633 @default.
• W4323306785 hasConceptScore W4323306785C52622490 @default.
• W4323306785 hasConceptScore W4323306785C62520636 @default.
• W4323306785 hasConceptScore W4323306785C76155785 @default.
• W4323306785 hasConceptScore W4323306785C94915269 @default.
• W4323306785 hasFunder F4320321001 @default.

• next