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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Cross-Domain Few-Shot Segmentation (CD-FSS) poses the challenge of segmentingnovel categories from a distinct domain using only limited exemplars. In thispaper, we undertake a comprehensive study of CD-FSS and uncover two crucialinsights: (i) the necessity of a fine-tuning stage to effectively transfer thelearned meta-knowledge across domains, and (ii) the overfitting risk during thena\"ive fine-tuning due to the scarcity of novel category examples. With theseinsights, we propose a novel cross-domain fine-tuning strategy that addressesthe challenging CD-FSS tasks. We first design Bi-directional Few-shotPrediction (BFP), which establishes support-query correspondence in abi-directional manner, crafting augmented supervision to reduce the overfittingrisk. Then we further extend BFP into Iterative Few-shot Adaptor (IFA), whichis a recursive framework to capture the support-query correspondenceiteratively, targeting maximal exploitation of supervisory signals from thesparse novel category samples. Extensive empirical evaluations show that ourmethod significantly outperforms the state-of-the-arts (+7.8\%), which verifiesthat IFA tackles the cross-domain challenges and mitigates the overfittingsimultaneously. Code will be made available.

Cross-Domain Few-Shot Segmentation via Iterative Support-Query Correspondence Mining