Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386598508> ?p ?o ?g. }
Showing items 1 to 65 of
65
with 100 items per page.
- W4386598508 abstract "Recent few-shot segmentation (FSS) methods introduce an extra pre-training stage before meta-training to obtain a stronger backbone, which has become a standard step in few-shot learning. Despite the effectiveness, current pre-training scheme suffers from the merged background problem: only base classes are labelled as foregrounds, making it hard to distinguish between novel classes and actual background. In this paper, we propose a new pre-training scheme for FSS via decoupling the novel classes from background, called Background Clustering Pre-Training (BCPT). Specifically, we adopt online clustering to the pixel embeddings of merged background to explore the underlying semantic structures, bridging the gap between pre-training and adaptation to novel classes. Given the clustering results, we further propose the background mining loss and leverage base classes to guide the clustering process, improving the quality and stability of clustering results. Experiments on PASCAL-5 <sup xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>i</sup> and COCO-20 <sup xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>i</sup> show that BCPT yields advanced performance. Code will be available at https://github.com/Carboxy/BCPT." @default.
- W4386598508 created "2023-09-12" @default.
- W4386598508 creator A5025097974 @default.
- W4386598508 creator A5081879869 @default.
- W4386598508 creator A5083440332 @default.
- W4386598508 date "2023-10-08" @default.
- W4386598508 modified "2023-09-30" @default.
- W4386598508 title "Background Clustering Pre-Training for Few-Shot Segmentation" @default.
- W4386598508 cites W2108598243 @default.
- W4386598508 cites W2194775991 @default.
- W4386598508 cites W2963078159 @default.
- W4386598508 cites W2981787211 @default.
- W4386598508 cites W2987741655 @default.
- W4386598508 cites W2994528761 @default.
- W4386598508 cites W3012255272 @default.
- W4386598508 cites W3034312118 @default.
- W4386598508 cites W3034576826 @default.
- W4386598508 cites W3093412994 @default.
- W4386598508 cites W3204077273 @default.
- W4386598508 cites W343636949 @default.
- W4386598508 cites W4214660208 @default.
- W4386598508 cites W4221153269 @default.
- W4386598508 cites W4224860128 @default.
- W4386598508 cites W4312592495 @default.
- W4386598508 cites W4312647976 @default.
- W4386598508 doi "https://doi.org/10.1109/icip49359.2023.10222055" @default.
- W4386598508 hasPublicationYear "2023" @default.
- W4386598508 type Work @default.
- W4386598508 citedByCount "0" @default.
- W4386598508 crossrefType "proceedings-article" @default.
- W4386598508 hasAuthorship W4386598508A5025097974 @default.
- W4386598508 hasAuthorship W4386598508A5081879869 @default.
- W4386598508 hasAuthorship W4386598508A5083440332 @default.
- W4386598508 hasBestOaLocation W43865985081 @default.
- W4386598508 hasConcept C119857082 @default.
- W4386598508 hasConcept C153083717 @default.
- W4386598508 hasConcept C153180895 @default.
- W4386598508 hasConcept C154945302 @default.
- W4386598508 hasConcept C41008148 @default.
- W4386598508 hasConcept C73555534 @default.
- W4386598508 hasConcept C89600930 @default.
- W4386598508 hasConceptScore W4386598508C119857082 @default.
- W4386598508 hasConceptScore W4386598508C153083717 @default.
- W4386598508 hasConceptScore W4386598508C153180895 @default.
- W4386598508 hasConceptScore W4386598508C154945302 @default.
- W4386598508 hasConceptScore W4386598508C41008148 @default.
- W4386598508 hasConceptScore W4386598508C73555534 @default.
- W4386598508 hasConceptScore W4386598508C89600930 @default.
- W4386598508 hasFunder F4320307804 @default.
- W4386598508 hasLocation W43865985081 @default.
- W4386598508 hasOpenAccess W4386598508 @default.
- W4386598508 hasPrimaryLocation W43865985081 @default.
- W4386598508 hasRelatedWork W2603501458 @default.
- W4386598508 hasRelatedWork W2961085424 @default.
- W4386598508 hasRelatedWork W3046775127 @default.
- W4386598508 hasRelatedWork W3170094116 @default.
- W4386598508 hasRelatedWork W4205958290 @default.
- W4386598508 hasRelatedWork W4285260836 @default.
- W4386598508 hasRelatedWork W4286629047 @default.
- W4386598508 hasRelatedWork W4306321456 @default.
- W4386598508 hasRelatedWork W4306674287 @default.
- W4386598508 hasRelatedWork W4224009465 @default.
- W4386598508 isParatext "false" @default.
- W4386598508 isRetracted "false" @default.
- W4386598508 workType "article" @default.