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

Image inpainting for completing complicated semantic environments and diversehole patterns of corrupted images is challenging even for state-of-the-artlearning-based inpainting methods trained on large-scale data. A referenceimage capturing the same scene of a corrupted image offers informative guidancefor completing the corrupted image as it shares similar texture and structurepriors to that of the holes of the corrupted image. In this work, we propose atransformer-based encoder-decoder network, named TransRef, for reference-guidedimage inpainting. Specifically, the guidance is conducted progressively througha reference embedding procedure, in which the referencing features aresubsequently aligned and fused with the features of the corrupted image. Forprecise utilization of the reference features for guidance, a reference-patchalignment (Ref-PA) module is proposed to align the patch features of thereference and corrupted images and harmonize their style differences, while areference-patch transformer (Ref-PT) module is proposed to refine the embeddedreference feature. Moreover, to facilitate the research of reference-guidedimage restoration tasks, we construct a publicly accessible benchmark datasetcontaining 50K pairs of input and reference images. Both quantitative andqualitative evaluations demonstrate the efficacy of the reference informationand the proposed method over the state-of-the-art methods in completing complexholes. Code and dataset can be accessed at https://github.com/Cameltr/TransRef.

TransRef: Multi-Scale Reference Embedding Transformer for Reference-Guided Image Inpainting