Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204261852> ?p ?o ?g. }
- W3204261852 abstract "We study self-supervised video representation learning, which is a challenging task due to 1) lack of labels for explicit supervision; 2) unstructured and noisy visual information. Existing methods mainly use contrastive loss with video clips as the instances and learn visual representation by discriminating instances from each other, but they need a careful treatment of negative pairs by either relying on large batch sizes, memory banks, extra modalities or customized mining strategies, which inevitably includes noisy data. In this paper, we observe that the consistency between positive samples is the key to learn robust video representation. Specifically, we propose two tasks to learn appearance and speed consistency, respectively. The appearance consistency task aims to maximize the similarity between two clips of the same video with different playback speeds. The speed consistency task aims to maximize the similarity between two clips with the same playback speed but different appearance information. We show that optimizing the two tasks jointly consistently improves the performance on downstream tasks, e.g., action recognition and video retrieval. Remarkably, for action recognition on the UCF-101 dataset, we achieve 90.8% accuracy without using any extra modalities or negative pairs for unsupervised pretraining, which outperforms the ImageNet supervised pretrained model. Codes and models will be available." @default.
- W3204261852 created "2021-10-11" @default.
- W3204261852 creator A5009613920 @default.
- W3204261852 creator A5015819673 @default.
- W3204261852 creator A5018276259 @default.
- W3204261852 creator A5024847477 @default.
- W3204261852 creator A5032352025 @default.
- W3204261852 creator A5035100425 @default.
- W3204261852 creator A5035916372 @default.
- W3204261852 creator A5050031109 @default.
- W3204261852 creator A5090639535 @default.
- W3204261852 date "2021-10-01" @default.
- W3204261852 modified "2023-10-05" @default.
- W3204261852 title "ASCNet: Self-supervised Video Representation Learning with Appearance-Speed Consistency" @default.
- W3204261852 cites W2126579184 @default.
- W3204261852 cites W2194775991 @default.
- W3204261852 cites W2487442924 @default.
- W3204261852 cites W2883451034 @default.
- W3204261852 cites W2948242301 @default.
- W3204261852 cites W2963155035 @default.
- W3204261852 cites W2963814513 @default.
- W3204261852 cites W2964037671 @default.
- W3204261852 cites W2997907976 @default.
- W3204261852 cites W3010010212 @default.
- W3204261852 cites W3010874390 @default.
- W3204261852 cites W3034215340 @default.
- W3204261852 cites W3034381931 @default.
- W3204261852 cites W3035524453 @default.
- W3204261852 cites W3104591054 @default.
- W3204261852 cites W3173166478 @default.
- W3204261852 cites W3174568846 @default.
- W3204261852 cites W3207340843 @default.
- W3204261852 doi "https://doi.org/10.1109/iccv48922.2021.00799" @default.
- W3204261852 hasPublicationYear "2021" @default.
- W3204261852 type Work @default.
- W3204261852 sameAs 3204261852 @default.
- W3204261852 citedByCount "22" @default.
- W3204261852 countsByYear W32042618522021 @default.
- W3204261852 countsByYear W32042618522022 @default.
- W3204261852 countsByYear W32042618522023 @default.
- W3204261852 crossrefType "proceedings-article" @default.
- W3204261852 hasAuthorship W3204261852A5009613920 @default.
- W3204261852 hasAuthorship W3204261852A5015819673 @default.
- W3204261852 hasAuthorship W3204261852A5018276259 @default.
- W3204261852 hasAuthorship W3204261852A5024847477 @default.
- W3204261852 hasAuthorship W3204261852A5032352025 @default.
- W3204261852 hasAuthorship W3204261852A5035100425 @default.
- W3204261852 hasAuthorship W3204261852A5035916372 @default.
- W3204261852 hasAuthorship W3204261852A5050031109 @default.
- W3204261852 hasAuthorship W3204261852A5090639535 @default.
- W3204261852 hasBestOaLocation W32042618522 @default.
- W3204261852 hasConcept C103278499 @default.
- W3204261852 hasConcept C115961682 @default.
- W3204261852 hasConcept C119857082 @default.
- W3204261852 hasConcept C144024400 @default.
- W3204261852 hasConcept C153180895 @default.
- W3204261852 hasConcept C154945302 @default.
- W3204261852 hasConcept C162324750 @default.
- W3204261852 hasConcept C17744445 @default.
- W3204261852 hasConcept C187736073 @default.
- W3204261852 hasConcept C199539241 @default.
- W3204261852 hasConcept C2776359362 @default.
- W3204261852 hasConcept C2776436953 @default.
- W3204261852 hasConcept C2778739407 @default.
- W3204261852 hasConcept C2779903281 @default.
- W3204261852 hasConcept C2780451532 @default.
- W3204261852 hasConcept C36289849 @default.
- W3204261852 hasConcept C41008148 @default.
- W3204261852 hasConcept C59404180 @default.
- W3204261852 hasConcept C94625758 @default.
- W3204261852 hasConceptScore W3204261852C103278499 @default.
- W3204261852 hasConceptScore W3204261852C115961682 @default.
- W3204261852 hasConceptScore W3204261852C119857082 @default.
- W3204261852 hasConceptScore W3204261852C144024400 @default.
- W3204261852 hasConceptScore W3204261852C153180895 @default.
- W3204261852 hasConceptScore W3204261852C154945302 @default.
- W3204261852 hasConceptScore W3204261852C162324750 @default.
- W3204261852 hasConceptScore W3204261852C17744445 @default.
- W3204261852 hasConceptScore W3204261852C187736073 @default.
- W3204261852 hasConceptScore W3204261852C199539241 @default.
- W3204261852 hasConceptScore W3204261852C2776359362 @default.
- W3204261852 hasConceptScore W3204261852C2776436953 @default.
- W3204261852 hasConceptScore W3204261852C2778739407 @default.
- W3204261852 hasConceptScore W3204261852C2779903281 @default.
- W3204261852 hasConceptScore W3204261852C2780451532 @default.
- W3204261852 hasConceptScore W3204261852C36289849 @default.
- W3204261852 hasConceptScore W3204261852C41008148 @default.
- W3204261852 hasConceptScore W3204261852C59404180 @default.
- W3204261852 hasConceptScore W3204261852C94625758 @default.
- W3204261852 hasFunder F4320318547 @default.
- W3204261852 hasFunder F4320321001 @default.
- W3204261852 hasFunder F4320322108 @default.
- W3204261852 hasLocation W32042618521 @default.
- W3204261852 hasLocation W32042618522 @default.
- W3204261852 hasOpenAccess W3204261852 @default.
- W3204261852 hasPrimaryLocation W32042618521 @default.
- W3204261852 hasRelatedWork W2292254049 @default.
- W3204261852 hasRelatedWork W2507989420 @default.
- W3204261852 hasRelatedWork W2546942002 @default.
- W3204261852 hasRelatedWork W2592385986 @default.