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

In this paper, we introduce and tackle the simultaneous enhancement andsuper-resolution (SESR) problem for underwater robot vision and provide anefficient solution for near real-time applications. We present Deep SESR, aresidual-in-residual network-based generative model that can learn to restoreperceptual image qualities at 2x, 3x, or 4x higher spatial resolution. Wesupervise its training by formulating a multi-modal objective function thataddresses the chrominance-specific underwater color degradation, lack of imagesharpness, and loss in high-level feature representation. It is also supervisedto learn salient foreground regions in the image, which in turn guides thenetwork to learn global contrast enhancement. We design an end-to-end trainingpipeline to jointly learn the saliency prediction and SESR on a sharedhierarchical feature space for fast inference. Moreover, we present UFO-120,the first dataset to facilitate large-scale SESR learning; it contains over1500 training samples and a benchmark test set of 120 samples. By thoroughexperimental evaluation on the UFO-120 and other standard datasets, wedemonstrate that Deep SESR outperforms the existing solutions for underwaterimage enhancement and super-resolution. We also validate its generalizationperformance on several test cases that include underwater images with diversespectral and spatial degradation levels, and also terrestrial images withunseen natural objects. Lastly, we analyze its computational feasibility forsingle-board deployments and demonstrate its operational benefits forvisually-guided underwater robots. The model and dataset information will beavailable at: https://github.com/xahidbuffon/Deep-SESR.

Simultaneous Enhancement and Super-Resolution of Underwater Imagery for Improved Visual Perception