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

Learning robust local image feature matching is a fundamental low-levelvision task, which has been widely explored in the past few years. Recently,detector-free local feature matchers based on transformers have shown promisingresults, which largely outperform pure Convolutional Neural Network (CNN) basedones. But correlations produced by transformer-based methods are spatiallylimited to the center of source views' coarse patches, because of the costlyattention learning. In this work, we rethink this issue and find that suchmatching formulation degrades pose estimation, especially for low-resolutionimages. So we propose a transformer-based cascade matching model -- Cascadefeature Matching TRansformer (CasMTR), to efficiently learn dense featurecorrelations, which allows us to choose more reliable matching pairs for therelative pose estimation. Instead of re-training a new detector, we use asimple yet effective Non-Maximum Suppression (NMS) post-process to filterkeypoints through the confidence map, and largely improve the matchingprecision. CasMTR achieves state-of-the-art performance in indoor and outdoorpose estimation as well as visual localization. Moreover, thorough ablationsshow the efficacy of the proposed components and techniques.

Improving Transformer-based Image Matching by Cascaded Capturing Spatially Informative Keypoints