Matches in SemOpenAlex for { <https://semopenalex.org/work/W4313306931> ?p ?o ?g. }
Showing items 1 to 84 of
84
with 100 items per page.
- W4313306931 abstract "Most existing unsupervised video hashing methods are built on unidirectional models with less effective similarity preservation objectives, which underuse the bidirectional correlations among successive frames and are susceptible to noisy data while constructing the similarity matrix. To alleviate the problems above, in this paper, we propose a novel unsupervised video hashing method, called Manifold Similarity Reconstruction based Video Hashing (MSVH). Inspired by the MLM (mask language model) architecture, we design a visual reconstruction loss to fully exploit the bidirectional correlations between frames by forcing the model to extract information from context. Different from the previous researches, a novel similarity reconstruction loss is adopt considering both Euclidean distance and manifold structure to cope with noise. Furthermore, we develop a more reliable classification loss to exploit the global statistics of the whole dataset as a complement to local similarity preservation. Extensive experiments implemented on three public benchmark datasets, FCVID, UCF101 and HMDB51, show that the proposed method outperforms the state-of-the-art baselines." @default.
- W4313306931 created "2023-01-06" @default.
- W4313306931 creator A5012286535 @default.
- W4313306931 creator A5061132796 @default.
- W4313306931 creator A5086837371 @default.
- W4313306931 date "2022-12-10" @default.
- W4313306931 modified "2023-09-23" @default.
- W4313306931 title "Manifold Similarity Reconstruction based Video Hashing" @default.
- W4313306931 cites W1777628566 @default.
- W4313306931 cites W1974647172 @default.
- W4313306931 cites W2108523493 @default.
- W4313306931 cites W2162006472 @default.
- W4313306931 cites W2165681080 @default.
- W4313306931 cites W2411707397 @default.
- W4313306931 cites W2527195341 @default.
- W4313306931 cites W2588651547 @default.
- W4313306931 cites W2786585376 @default.
- W4313306931 cites W2900790473 @default.
- W4313306931 cites W2981851019 @default.
- W4313306931 cites W2996754048 @default.
- W4313306931 cites W3034205626 @default.
- W4313306931 cites W3043840704 @default.
- W4313306931 cites W3177244733 @default.
- W4313306931 doi "https://doi.org/10.1109/hdis56859.2022.9991567" @default.
- W4313306931 hasPublicationYear "2022" @default.
- W4313306931 type Work @default.
- W4313306931 citedByCount "0" @default.
- W4313306931 crossrefType "proceedings-article" @default.
- W4313306931 hasAuthorship W4313306931A5012286535 @default.
- W4313306931 hasAuthorship W4313306931A5061132796 @default.
- W4313306931 hasAuthorship W4313306931A5086837371 @default.
- W4313306931 hasConcept C103278499 @default.
- W4313306931 hasConcept C115961682 @default.
- W4313306931 hasConcept C116738811 @default.
- W4313306931 hasConcept C119857082 @default.
- W4313306931 hasConcept C120174047 @default.
- W4313306931 hasConcept C124101348 @default.
- W4313306931 hasConcept C13280743 @default.
- W4313306931 hasConcept C151730666 @default.
- W4313306931 hasConcept C153180895 @default.
- W4313306931 hasConcept C154945302 @default.
- W4313306931 hasConcept C165696696 @default.
- W4313306931 hasConcept C185798385 @default.
- W4313306931 hasConcept C205649164 @default.
- W4313306931 hasConcept C2779343474 @default.
- W4313306931 hasConcept C38652104 @default.
- W4313306931 hasConcept C41008148 @default.
- W4313306931 hasConcept C86803240 @default.
- W4313306931 hasConcept C99138194 @default.
- W4313306931 hasConceptScore W4313306931C103278499 @default.
- W4313306931 hasConceptScore W4313306931C115961682 @default.
- W4313306931 hasConceptScore W4313306931C116738811 @default.
- W4313306931 hasConceptScore W4313306931C119857082 @default.
- W4313306931 hasConceptScore W4313306931C120174047 @default.
- W4313306931 hasConceptScore W4313306931C124101348 @default.
- W4313306931 hasConceptScore W4313306931C13280743 @default.
- W4313306931 hasConceptScore W4313306931C151730666 @default.
- W4313306931 hasConceptScore W4313306931C153180895 @default.
- W4313306931 hasConceptScore W4313306931C154945302 @default.
- W4313306931 hasConceptScore W4313306931C165696696 @default.
- W4313306931 hasConceptScore W4313306931C185798385 @default.
- W4313306931 hasConceptScore W4313306931C205649164 @default.
- W4313306931 hasConceptScore W4313306931C2779343474 @default.
- W4313306931 hasConceptScore W4313306931C38652104 @default.
- W4313306931 hasConceptScore W4313306931C41008148 @default.
- W4313306931 hasConceptScore W4313306931C86803240 @default.
- W4313306931 hasConceptScore W4313306931C99138194 @default.
- W4313306931 hasFunder F4320321001 @default.
- W4313306931 hasLocation W43133069311 @default.
- W4313306931 hasOpenAccess W4313306931 @default.
- W4313306931 hasPrimaryLocation W43133069311 @default.
- W4313306931 hasRelatedWork W2015538044 @default.
- W4313306931 hasRelatedWork W2560193952 @default.
- W4313306931 hasRelatedWork W2588629527 @default.
- W4313306931 hasRelatedWork W2810523766 @default.
- W4313306931 hasRelatedWork W3027599074 @default.
- W4313306931 hasRelatedWork W3029685757 @default.
- W4313306931 hasRelatedWork W3131368048 @default.
- W4313306931 hasRelatedWork W3161472561 @default.
- W4313306931 hasRelatedWork W3183633970 @default.
- W4313306931 hasRelatedWork W4289653936 @default.
- W4313306931 isParatext "false" @default.
- W4313306931 isRetracted "false" @default.
- W4313306931 workType "article" @default.