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

We present a novel deep learning approach to synthesize complete face imagesin the presence of large ocular region occlusions. This is motivated by recentsurge of VR/AR displays that hinder face-to-face communications. Different fromthe state-of-the-art face inpainting methods that have no control over thesynthesized content and can only handle frontal face pose, our approach canfaithfully recover the missing content under various head poses whilepreserving the identity. At the core of our method is a novel generativenetwork with dedicated constraints to regularize the synthesis process. Topreserve the identity, our network takes an arbitrary occlusion-free image ofthe target identity to infer the missing content, and its high-level CNNfeatures as an identity prior to regularize the searching space of generator.Since the input reference image may have a different pose, a pose map and anovel pose discriminator are further adopted to supervise the learning ofimplicit pose transformations. Our method is capable of generating coherentfacial inpainting with consistent identity over videos with large variations ofhead motions. Experiments on both synthesized and real data demonstrate thatour method greatly outperforms the state-of-the-art methods in terms of bothsynthesis quality and robustness.

Identity Preserving Face Completion for Large Ocular Region Occlusion