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

With the advent of convolutional neural networks, stereo matching algorithmshave recently gained tremendous progress. However, it remains a great challengeto accurately extract disparities from real-world image pairs taken byconsumer-level devices like smartphones, due to practical complicating factorssuch as thin structures, non-ideal rectification, camera module inconsistenciesand various hard-case scenes. In this paper, we propose a set of innovativedesigns to tackle the problem of practical stereo matching: 1) to betterrecover fine depth details, we design a hierarchical network with recurrentrefinement to update disparities in a coarse-to-fine manner, as well as astacked cascaded architecture for inference; 2) we propose an adaptive groupcorrelation layer to mitigate the impact of erroneous rectification; 3) weintroduce a new synthetic dataset with special attention to difficult cases forbetter generalizing to real-world scenes. Our results not only rank 1st on bothMiddlebury and ETH3D benchmarks, outperforming existing state-of-the-artmethods by a notable margin, but also exhibit high-quality details forreal-life photos, which clearly demonstrates the efficacy of our contributions.

Practical Stereo Matching via Cascaded Recurrent Network with Adaptive Correlation