Matches in SemOpenAlex for { <https://semopenalex.org/work/W3203649342> ?p ?o ?g. }
- W3203649342 abstract "We focus on the problem of segmenting a certain object referred by a natural language sentence in video content, at the core of formulating a pinpoint vision-language relation. While existing attempts mainly construct such relation in an implicit way, i.e., grid-level multi-modal feature fusion, it has been proven problematic to distinguish semantically similar objects under this paradigm. In this work, we propose to interwind the visual and linguistic modalities in an explicit way via the contrastive learning objective, which directly aligns the referred object and the language description and separates the unreferred content apart across frames. Moreover, to remedy for the degradation problem, we present two complementary hard instance mining strategies, i.e., Language-relevant Channel Filter and Relative Hard Instance Construction. They encourage the network to exclude visual-distinguishable feature and to focus on easy-confused objects during the contrastive training. Extensive experiments on two benchmarks, i.e., A2D Sentences and J-HMDB Sentences, quantitatively demonstrate the state-of-the-arts performance of our method and qualitatively show the more accurate distinguishment between semantically similar objects over baselines." @default.
- W3203649342 created "2021-10-11" @default.
- W3203649342 creator A5010961909 @default.
- W3203649342 creator A5045556326 @default.
- W3203649342 creator A5071501764 @default.
- W3203649342 creator A5072690470 @default.
- W3203649342 date "2021-09-28" @default.
- W3203649342 modified "2023-10-14" @default.
- W3203649342 title "Contrastive Video-Language Segmentation" @default.
- W3203649342 cites W1849277567 @default.
- W3203649342 cites W1905722737 @default.
- W3203649342 cites W1921523184 @default.
- W3203649342 cites W1940872118 @default.
- W3203649342 cites W2034014085 @default.
- W3203649342 cites W2108598243 @default.
- W3203649342 cites W2144796873 @default.
- W3203649342 cites W2194775991 @default.
- W3203649342 cites W2251512949 @default.
- W3203649342 cites W2302548814 @default.
- W3203649342 cites W2550553598 @default.
- W3203649342 cites W2558661413 @default.
- W3203649342 cites W2747053578 @default.
- W3203649342 cites W2798991696 @default.
- W3203649342 cites W2910628332 @default.
- W3203649342 cites W2949517790 @default.
- W3203649342 cites W2963354481 @default.
- W3203649342 cites W2963420686 @default.
- W3203649342 cites W2964121744 @default.
- W3203649342 cites W2983693499 @default.
- W3203649342 cites W2990113535 @default.
- W3203649342 cites W2997063389 @default.
- W3203649342 cites W3010526025 @default.
- W3203649342 cites W3032554070 @default.
- W3203649342 cites W3034777757 @default.
- W3203649342 cites W3034804856 @default.
- W3203649342 cites W3034978746 @default.
- W3203649342 cites W3035524453 @default.
- W3203649342 cites W3091546937 @default.
- W3203649342 cites W3113328489 @default.
- W3203649342 cites W3117585461 @default.
- W3203649342 cites W3122924117 @default.
- W3203649342 cites W3139267983 @default.
- W3203649342 cites W3171516518 @default.
- W3203649342 cites W3175335326 @default.
- W3203649342 cites W3177892185 @default.
- W3203649342 cites W3182236906 @default.
- W3203649342 cites W3202931577 @default.
- W3203649342 cites W343636949 @default.
- W3203649342 doi "https://doi.org/10.48550/arxiv.2109.14131" @default.
- W3203649342 hasPublicationYear "2021" @default.
- W3203649342 type Work @default.
- W3203649342 sameAs 3203649342 @default.
- W3203649342 citedByCount "0" @default.
- W3203649342 crossrefType "posted-content" @default.
- W3203649342 hasAuthorship W3203649342A5010961909 @default.
- W3203649342 hasAuthorship W3203649342A5045556326 @default.
- W3203649342 hasAuthorship W3203649342A5071501764 @default.
- W3203649342 hasAuthorship W3203649342A5072690470 @default.
- W3203649342 hasBestOaLocation W32036493421 @default.
- W3203649342 hasConcept C106131492 @default.
- W3203649342 hasConcept C120665830 @default.
- W3203649342 hasConcept C121332964 @default.
- W3203649342 hasConcept C138885662 @default.
- W3203649342 hasConcept C154945302 @default.
- W3203649342 hasConcept C185592680 @default.
- W3203649342 hasConcept C188027245 @default.
- W3203649342 hasConcept C192209626 @default.
- W3203649342 hasConcept C195324797 @default.
- W3203649342 hasConcept C199360897 @default.
- W3203649342 hasConcept C204321447 @default.
- W3203649342 hasConcept C25343380 @default.
- W3203649342 hasConcept C2776401178 @default.
- W3203649342 hasConcept C2777530160 @default.
- W3203649342 hasConcept C2780801425 @default.
- W3203649342 hasConcept C2781238097 @default.
- W3203649342 hasConcept C31972630 @default.
- W3203649342 hasConcept C41008148 @default.
- W3203649342 hasConcept C41895202 @default.
- W3203649342 hasConcept C71139939 @default.
- W3203649342 hasConcept C77088390 @default.
- W3203649342 hasConcept C89600930 @default.
- W3203649342 hasConceptScore W3203649342C106131492 @default.
- W3203649342 hasConceptScore W3203649342C120665830 @default.
- W3203649342 hasConceptScore W3203649342C121332964 @default.
- W3203649342 hasConceptScore W3203649342C138885662 @default.
- W3203649342 hasConceptScore W3203649342C154945302 @default.
- W3203649342 hasConceptScore W3203649342C185592680 @default.
- W3203649342 hasConceptScore W3203649342C188027245 @default.
- W3203649342 hasConceptScore W3203649342C192209626 @default.
- W3203649342 hasConceptScore W3203649342C195324797 @default.
- W3203649342 hasConceptScore W3203649342C199360897 @default.
- W3203649342 hasConceptScore W3203649342C204321447 @default.
- W3203649342 hasConceptScore W3203649342C25343380 @default.
- W3203649342 hasConceptScore W3203649342C2776401178 @default.
- W3203649342 hasConceptScore W3203649342C2777530160 @default.
- W3203649342 hasConceptScore W3203649342C2780801425 @default.
- W3203649342 hasConceptScore W3203649342C2781238097 @default.
- W3203649342 hasConceptScore W3203649342C31972630 @default.
- W3203649342 hasConceptScore W3203649342C41008148 @default.
- W3203649342 hasConceptScore W3203649342C41895202 @default.