DC Field | Value | Language |
---|---|---|
dc.contributor.author | JU, HONG MIN | - |
dc.contributor.author | DAHYUN, KANG | - |
dc.contributor.author | CHO, MINSU | - |
dc.date.accessioned | 2021-12-05T11:25:36Z | - |
dc.date.available | 2021-12-05T11:25:36Z | - |
dc.date.created | 2021-11-24 | - |
dc.date.issued | 2021-10-12 | - |
dc.identifier.uri | https://oasis.postech.ac.kr/handle/2014.oak/108185 | - |
dc.description.abstract | Few-shot semantic segmentation aims at learning to segment a target object from a query image using only a few annotated support images of the target class. This challenging task requires to understand diverse levels of visual cues and analyze fine-grained correspondence relations between the query and the support images. To address the problem, we propose Hypercorrelation Squeeze Networks (HSNet) that leverages multi-level feature correlation and efficient 4D convolutions. It extracts diverse features from different levels of intermediate convolutional layers and constructs a collection of 4D correlation tensors, i.e., hypercorrelations. Using efficient center-pivot 4D convolutions in a pyramidal architecture, the method gradually squeezes high-level semantic and low-level geometric cues of the hypercorrelation into precise segmentation masks in coarse-to-fine manner. The significant performance improvements on standard few-shot segmentation benchmarks of PASCAL-5i, COCO-20i, and FSS-1000 verify the efficacy of the proposed method. | - |
dc.publisher | IEEE / CVF | - |
dc.relation.isPartOf | International Conference on Computer Vision | - |
dc.title | Hypercorrelation Squeeze for Few-Shot Segmentation | - |
dc.type | Conference | - |
dc.type.rims | CONF | - |
dc.identifier.bibliographicCitation | International Conference on Computer Vision | - |
dc.citation.conferenceDate | 2021-10-11 | - |
dc.citation.conferencePlace | CN | - |
dc.citation.title | International Conference on Computer Vision | - |
dc.contributor.affiliatedAuthor | JU, HONG MIN | - |
dc.contributor.affiliatedAuthor | DAHYUN, KANG | - |
dc.contributor.affiliatedAuthor | CHO, MINSU | - |
dc.description.journalClass | 1 | - |
dc.description.journalClass | 1 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.
library@postech.ac.kr Tel: 054-279-2548
Copyrights © by 2017 Pohang University of Science ad Technology All right reserved.