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

Weakly-supervised image segmentation (WSIS) is a critical task in computervision that relies on image-level class labels. Multi-stage training procedureshave been widely used in existing WSIS approaches to obtain high-qualitypseudo-masks as ground-truth, resulting in significant progress. However,single-stage WSIS methods have recently gained attention due to their potentialfor simplifying training procedures, despite often suffering from low-qualitypseudo-masks that limit their practical applications. To address this issue, wepropose a novel single-stage WSIS method that utilizes a siamese network withcontrastive learning to improve the quality of class activation maps (CAMs) andachieve a self-refinement process. Our approach employs a cross-representationrefinement method that expands reliable object regions by utilizing differentfeature representations from the backbone. Additionally, we introduce across-transform regularization module that learns robust class prototypes forcontrastive learning and captures global context information to feed back roughCAMs, thereby improving the quality of CAMs. Our final high-quality CAMs areused as pseudo-masks to supervise the segmentation result. Experimental resultson the PASCAL VOC 2012 dataset demonstrate that our method significantlyoutperforms other state-of-the-art methods, achieving 67.2% and 68.76% mIoU onPASCAL VOC 2012 val set and test set, respectively. Furthermore, our method hasbeen extended to weakly supervised object localization task, and experimentalresults demonstrate that our method continues to achieve very competitiveresults.

Boosting Weakly-Supervised Image Segmentation via Representation, Transform, and Compensator