FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386071527> ?p ?o ?g. }

• W4386071527 abstract "The task of weakly supervised temporal action localization targets at generating temporal boundaries for actions of interest, meanwhile the action category should also be classified. Pseudo-label-based methods, which serve as an effective solution, have been widely studied recently. However, existing methods generate pseudo labels during training and make predictions during testing under different pipelines or settings, resulting in a gap between training and testing. In this paper, we propose to generate high-quality pseudo labels from the predicted action boundaries. Nevertheless, we note that existing post-processing, like NMS, would lead to information loss, which is insufficient to generate high-quality action boundaries. More importantly, transforming action boundaries into pseudo labels is quite challenging, since the predicted action instances are generally overlapped and have different confidence scores. Besides, the generated pseudo-labels can be fluctuating and inaccurate at the early stage of training. It might repeatedly strengthen the false predictions if there is no mechanism to conduct self-correction. To tackle these issues, we come up with an effective pipeline for learning better pseudo labels. Firstly, we propose a Gaussian weighted fusion module to preserve information of action instances and obtain high-quality action boundaries. Second, we formulate the pseudo-label generation as an optimization problem under the constraints in terms of the confidence scores of action instances. Finally, we introduce the idea of <tex xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>$Delta$</tex> pseudo labels, which enables the model with the ability of self-correction. Our method achieves superior performance to existing methods on two benchmarks, THUMOS14 and ActivityNet1.3, achieving gains of 1.9% on THUMOS14 and 3.7% on ActivityNet1.3 in terms of average mAP. Our code is available at https://github.com/zhou745/GauFuse_WSTAL.git." @default.
• W4386071527 created "2023-08-23" @default.
• W4386071527 creator A5002276547 @default.
• W4386071527 creator A5008232261 @default.
• W4386071527 creator A5047527718 @default.
• W4386071527 creator A5065073978 @default.
• W4386071527 creator A5086664647 @default.
• W4386071527 date "2023-06-01" @default.
• W4386071527 modified "2023-09-27" @default.
• W4386071527 title "Improving Weakly Supervised Temporal Action Localization by Bridging Train-Test Gap in Pseudo Labels" @default.
• W4386071527 cites W1927052826 @default.
• W4386071527 cites W2146444479 @default.
• W4386071527 cites W2604113307 @default.
• W4386071527 cites W2948229620 @default.
• W4386071527 cites W2952435096 @default.
• W4386071527 cites W2962709777 @default.
• W4386071527 cites W2962728572 @default.
• W4386071527 cites W2962876901 @default.
• W4386071527 cites W2963314614 @default.
• W4386071527 cites W2963524571 @default.
• W4386071527 cites W2964216549 @default.
• W4386071527 cites W2964274719 @default.
• W4386071527 cites W2983918066 @default.
• W4386071527 cites W2984478308 @default.
• W4386071527 cites W2984619425 @default.
• W4386071527 cites W2988098865 @default.
• W4386071527 cites W2998601171 @default.
• W4386071527 cites W2998702159 @default.
• W4386071527 cites W3092976396 @default.
• W4386071527 cites W3099206234 @default.
• W4386071527 cites W3112040205 @default.
• W4386071527 cites W3118479201 @default.
• W4386071527 cites W3173212682 @default.
• W4386071527 cites W3173698268 @default.
• W4386071527 cites W3173874725 @default.
• W4386071527 cites W3174421047 @default.
• W4386071527 cites W3174511093 @default.
• W4386071527 cites W3175580044 @default.
• W4386071527 cites W3176853098 @default.
• W4386071527 cites W3178036208 @default.
• W4386071527 cites W3189800722 @default.
• W4386071527 cites W3203848195 @default.
• W4386071527 cites W3207927851 @default.
• W4386071527 cites W4226500165 @default.
• W4386071527 cites W4301963599 @default.
• W4386071527 cites W4304080207 @default.
• W4386071527 doi "https://doi.org/10.1109/cvpr52729.2023.02203" @default.
• W4386071527 hasPublicationYear "2023" @default.
• W4386071527 type Work @default.
• W4386071527 citedByCount "0" @default.
• W4386071527 crossrefType "proceedings-article" @default.
• W4386071527 hasAuthorship W4386071527A5002276547 @default.
• W4386071527 hasAuthorship W4386071527A5008232261 @default.
• W4386071527 hasAuthorship W4386071527A5047527718 @default.
• W4386071527 hasAuthorship W4386071527A5065073978 @default.
• W4386071527 hasAuthorship W4386071527A5086664647 @default.
• W4386071527 hasConcept C111472728 @default.
• W4386071527 hasConcept C119857082 @default.
• W4386071527 hasConcept C121332964 @default.
• W4386071527 hasConcept C138885662 @default.
• W4386071527 hasConcept C154945302 @default.
• W4386071527 hasConcept C162324750 @default.
• W4386071527 hasConcept C174348530 @default.
• W4386071527 hasConcept C187736073 @default.
• W4386071527 hasConcept C199360897 @default.
• W4386071527 hasConcept C2779530757 @default.
• W4386071527 hasConcept C2780451532 @default.
• W4386071527 hasConcept C2780791683 @default.
• W4386071527 hasConcept C31258907 @default.
• W4386071527 hasConcept C41008148 @default.
• W4386071527 hasConcept C43521106 @default.
• W4386071527 hasConcept C62520636 @default.
• W4386071527 hasConceptScore W4386071527C111472728 @default.
• W4386071527 hasConceptScore W4386071527C119857082 @default.
• W4386071527 hasConceptScore W4386071527C121332964 @default.
• W4386071527 hasConceptScore W4386071527C138885662 @default.
• W4386071527 hasConceptScore W4386071527C154945302 @default.
• W4386071527 hasConceptScore W4386071527C162324750 @default.
• W4386071527 hasConceptScore W4386071527C174348530 @default.
• W4386071527 hasConceptScore W4386071527C187736073 @default.
• W4386071527 hasConceptScore W4386071527C199360897 @default.
• W4386071527 hasConceptScore W4386071527C2779530757 @default.
• W4386071527 hasConceptScore W4386071527C2780451532 @default.
• W4386071527 hasConceptScore W4386071527C2780791683 @default.
• W4386071527 hasConceptScore W4386071527C31258907 @default.
• W4386071527 hasConceptScore W4386071527C41008148 @default.
• W4386071527 hasConceptScore W4386071527C43521106 @default.
• W4386071527 hasConceptScore W4386071527C62520636 @default.
• W4386071527 hasFunder F4320335777 @default.
• W4386071527 hasLocation W43860715271 @default.
• W4386071527 hasOpenAccess W4386071527 @default.
• W4386071527 hasPrimaryLocation W43860715271 @default.
• W4386071527 hasRelatedWork W1807477017 @default.
• W4386071527 hasRelatedWork W2081647779 @default.
• W4386071527 hasRelatedWork W2961085424 @default.
• W4386071527 hasRelatedWork W2992516105 @default.
• W4386071527 hasRelatedWork W3046775127 @default.
• W4386071527 hasRelatedWork W4285260836 @default.
• W4386071527 hasRelatedWork W4286629047 @default.
• W4386071527 hasRelatedWork W4306321456 @default.

• next