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

Learning-based gaze estimation methods require large amounts of training datawith accurate gaze annotations. Facing such demanding requirements of gaze datacollection and annotation, several image synthesis methods were proposed, whichsuccessfully redirected gaze directions precisely given the assignedconditions. However, these methods focused on changing gaze directions of theimages that only include eyes or restricted ranges of faces with low resolution(less than $128\times128$) to largely reduce interference from other attributessuch as hairs, which limits application scenarios. To cope with thislimitation, we proposed a portable network, called ReDirTrans, achievinglatent-to-latent translation for redirecting gaze directions and headorientations in an interpretable manner. ReDirTrans projects input latentvectors into aimed-attribute embeddings only and redirects these embeddingswith assigned pitch and yaw values. Then both the initial and edited embeddingsare projected back (deprojected) to the initial latent space as residuals tomodify the input latent vectors by subtraction and addition, representing oldstatus removal and new status addition. The projection of aimed attributes onlyand subtraction-addition operations for status replacement essentially mitigateimpacts on other attributes and the distribution of latent vectors. Thus, bycombining ReDirTrans with a pretrained fixed e4e-StyleGAN pair, we createdReDirTrans-GAN, which enables accurately redirecting gaze in full-face imageswith $1024\times1024$ resolution while preserving other attributes such asidentity, expression, and hairstyle. Furthermore, we presented improvements forthe downstream learning-based gaze estimation task, using redirected samples asdataset augmentation.

ReDirTrans: Latent-to-Latent Translation for Gaze and Head Redirection