English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Noisy labels, inevitably existing in pseudo segmentation labels generatedfrom weak object-level annotations, severely hampers model optimization forsemantic segmentation. Previous works often rely on massive hand-crafted lossesand carefully-tuned hyper-parameters to resist noise, suffering poorgeneralization capability and high model complexity. Inspired by recentadvances in meta learning, we argue that rather than struggling to toleratenoise hidden behind clean labels passively, a more feasible solution would beto find out the noisy regions actively, so as to simply ignore them duringmodel optimization. With this in mind, this work presents a novel meta learningbased semantic segmentation method, MetaSeg, that comprises a primarycontent-aware meta-net (CAM-Net) to sever as a noise indicator for an arbitrarysegmentation model counterpart. Specifically, CAM-Net learns to generatepixel-wise weights to suppress noisy regions with incorrect pseudo labels whilehighlighting clean ones by exploiting hybrid strengthened features from imagecontent, providing straightforward and reliable guidance for optimizing thesegmentation model. Moreover, to break the barrier of time-consuming trainingwhen applying meta learning to common large segmentation models, we furtherpresent a new decoupled training strategy that optimizes different model layersin a divide-and-conquer manner. Extensive experiments on object, medical,remote sensing and human segmentation shows that our method achieves superiorperformance, approaching that of fully supervised settings, which paves a newpromising way for omni-supervised semantic segmentation.

MetaSeg: Content-Aware Meta-Net for Omni-Supervised Semantic Segmentation