FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4313010522> ?p ?o ?g. }

• W4313010522 endingPage "77" @default.
• W4313010522 startingPage "59" @default.
• W4313010522 abstract "To improve instance-level detection/segmentation performance, existing self-supervised and semi-supervised methods extract either task-unrelated or task-specific training signals from unlabeled data. We show that these two approaches, at the two extreme ends of the task-specificity spectrum, are suboptimal for the task performance. Utilizing too little task-specific training signals causes underfitting to the ground-truth labels of downstream tasks, while the opposite causes overfitting to the ground-truth labels. To this end, we propose a novel Class-Agnostic Semi-Supervised Learning (CA-SSL) framework to achieve a more favorable task-specificity balance in extracting training signals from unlabeled data. CA-SSL has three training stages that act on either ground-truth labels (labeled data) or pseudo labels (unlabeled data). This decoupling strategy avoids the complicated scheme in traditional SSL methods that balances the contributions from both data types. Especially, we introduce a warmup training stage to achieve a more optimal balance in task specificity by ignoring class information in the pseudo labels, while preserving localization training signals. As a result, our warmup model can better avoid underfitting/overfitting when fine-tuned on the ground-truth labels in detection and segmentation tasks. Using 3.6M unlabeled data, we achieve a significant performance gain of $$4.7%$$ over ImageNet-pretrained baseline on FCOS object detection. In addition, our warmup model demonstrates excellent transferability to other detection and segmentation frameworks." @default.
• W4313010522 created "2023-01-05" @default.
• W4313010522 creator A5005119482 @default.
• W4313010522 creator A5018635864 @default.
• W4313010522 creator A5024456511 @default.
• W4313010522 creator A5037049317 @default.
• W4313010522 creator A5045358378 @default.
• W4313010522 creator A5052856441 @default.
• W4313010522 creator A5063179713 @default.
• W4313010522 creator A5076304703 @default.
• W4313010522 creator A5076366439 @default.
• W4313010522 creator A5087826010 @default.
• W4313010522 date "2022-01-01" @default.
• W4313010522 modified "2023-10-16" @default.
• W4313010522 title "CA-SSL: Class-Agnostic Semi-Supervised Learning for Detection and Segmentation" @default.
• W4313010522 cites W1861492603 @default.
• W4313010522 cites W1976047850 @default.
• W4313010522 cites W2097117768 @default.
• W4313010522 cites W2102605133 @default.
• W4313010522 cites W2108598243 @default.
• W4313010522 cites W2125215748 @default.
• W4313010522 cites W2183341477 @default.
• W4313010522 cites W2194775991 @default.
• W4313010522 cites W2340897893 @default.
• W4313010522 cites W2565639579 @default.
• W4313010522 cites W2732026016 @default.
• W4313010522 cites W2752782242 @default.
• W4313010522 cites W2910628332 @default.
• W4313010522 cites W2916798096 @default.
• W4313010522 cites W2963150697 @default.
• W4313010522 cites W2963296620 @default.
• W4313010522 cites W2963351448 @default.
• W4313010522 cites W2963394817 @default.
• W4313010522 cites W2963402313 @default.
• W4313010522 cites W2963446712 @default.
• W4313010522 cites W2963785012 @default.
• W4313010522 cites W2963849369 @default.
• W4313010522 cites W2963857746 @default.
• W4313010522 cites W2964325922 @default.
• W4313010522 cites W2964350391 @default.
• W4313010522 cites W2982161360 @default.
• W4313010522 cites W2982770724 @default.
• W4313010522 cites W2991522752 @default.
• W4313010522 cites W3034428102 @default.
• W4313010522 cites W3034826836 @default.
• W4313010522 cites W3035524453 @default.
• W4313010522 cites W3096609285 @default.
• W4313010522 cites W3097722521 @default.
• W4313010522 cites W3106546328 @default.
• W4313010522 cites W3120055985 @default.
• W4313010522 cites W3128661784 @default.
• W4313010522 cites W3138516171 @default.
• W4313010522 cites W3158120491 @default.
• W4313010522 cites W3165312683 @default.
• W4313010522 cites W3165924482 @default.
• W4313010522 cites W3168822201 @default.
• W4313010522 cites W3170410843 @default.
• W4313010522 cites W3171007011 @default.
• W4313010522 cites W3172507542 @default.
• W4313010522 cites W3175086476 @default.
• W4313010522 cites W3184021524 @default.
• W4313010522 cites W3194038966 @default.
• W4313010522 cites W3203373254 @default.
• W4313010522 cites W3214129952 @default.
• W4313010522 cites W4214504406 @default.
• W4313010522 cites W4288083516 @default.
• W4313010522 cites W4313176857 @default.
• W4313010522 doi "https://doi.org/10.1007/978-3-031-19821-2_4" @default.
• W4313010522 hasPublicationYear "2022" @default.
• W4313010522 type Work @default.
• W4313010522 citedByCount "0" @default.
• W4313010522 crossrefType "book-chapter" @default.
• W4313010522 hasAuthorship W4313010522A5005119482 @default.
• W4313010522 hasAuthorship W4313010522A5018635864 @default.
• W4313010522 hasAuthorship W4313010522A5024456511 @default.
• W4313010522 hasAuthorship W4313010522A5037049317 @default.
• W4313010522 hasAuthorship W4313010522A5045358378 @default.
• W4313010522 hasAuthorship W4313010522A5052856441 @default.
• W4313010522 hasAuthorship W4313010522A5063179713 @default.
• W4313010522 hasAuthorship W4313010522A5076304703 @default.
• W4313010522 hasAuthorship W4313010522A5076366439 @default.
• W4313010522 hasAuthorship W4313010522A5087826010 @default.
• W4313010522 hasBestOaLocation W43130105222 @default.
• W4313010522 hasConcept C119857082 @default.
• W4313010522 hasConcept C146849305 @default.
• W4313010522 hasConcept C153180895 @default.
• W4313010522 hasConcept C154945302 @default.
• W4313010522 hasConcept C162324750 @default.
• W4313010522 hasConcept C187736073 @default.
• W4313010522 hasConcept C22019652 @default.
• W4313010522 hasConcept C2776145971 @default.
• W4313010522 hasConcept C2777212361 @default.
• W4313010522 hasConcept C2780451532 @default.
• W4313010522 hasConcept C41008148 @default.
• W4313010522 hasConcept C50644808 @default.
• W4313010522 hasConcept C89600930 @default.
• W4313010522 hasConceptScore W4313010522C119857082 @default.
• W4313010522 hasConceptScore W4313010522C146849305 @default.

• next