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

Recent studies of two-view correspondence learning usually establish anend-to-end network to jointly predict correspondence reliability and relativepose. We improve such a framework from two aspects. First, we propose a LocalFeature Consensus (LFC) plugin block to augment the features of existingmodels. Given a correspondence feature, the block augments its neighboringfeatures with mutual neighborhood consensus and aggregates them to produce anenhanced feature. As inliers obey a uniform cross-view transformation and sharemore consistent learned features than outliers, feature consensus strengthensinlier correlation and suppresses outlier distraction, which makes outputfeatures more discriminative for classifying inliers/outliers. Second, existingapproaches supervise network training with the ground truth correspondences andessential matrix projecting one image to the other for an input image pair,without considering the information from the reverse mapping. We extendexisting models to a Siamese network with a reciprocal loss that exploits thesupervision of mutual projection, which considerably promotes the matchingperformance without introducing additional model parameters. Building uponMSA-Net, we implement the two proposals and experimentally achievestate-of-the-art performance on benchmark datasets.

Local Consensus Enhanced Siamese Network with Reciprocal Loss for Two-view Correspondence Learning