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

Data augmentation is an effective way to improve the performance of deepnetworks. Unfortunately, current methods are mostly developed for high-levelvision tasks (e.g., classification) and few are studied for low-level visiontasks (e.g., image restoration). In this paper, we provide a comprehensiveanalysis of the existing augmentation methods applied to the super-resolutiontask. We find that the methods discarding or manipulating the pixels orfeatures too much hamper the image restoration, where the spatial relationshipis very important. Based on our analyses, we propose CutBlur that cuts alow-resolution patch and pastes it to the corresponding high-resolution imageregion and vice versa. The key intuition of CutBlur is to enable a model tolearn not only "how" but also "where" to super-resolve an image. By doing so,the model can understand "how much", instead of blindly learning to applysuper-resolution to every given pixel. Our method consistently andsignificantly improves the performance across various scenarios, especiallywhen the model size is big and the data is collected under real-worldenvironments. We also show that our method improves other low-level visiontasks, such as denoising and compression artifact removal.

Rethinking Data Augmentation for Image Super-resolution: A Comprehensive Analysis and a New Strategy