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/

Zendy Plus

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

Zendy Tools

Zendy Open

Recently, manipulation of real-world images has been highly elaborated alongwith the development of Generative Adversarial Networks (GANs) andcorresponding encoders, which embed real-world images into the latent space.However, designing encoders of GAN still remains a challenging task due to thetrade-off between distortion and perception. In this paper, we point out thatthe existing encoders try to lower the distortion not only on the interestregion, e.g., human facial region but also on the uninterest region, e.g.,background patterns and obstacles. However, most uninterest regions inreal-world images are located at out-of-distribution (OOD), which areinfeasible to be ideally reconstructed by generative models. Moreover, weempirically find that the uninterest region overlapped with the interest regioncan mangle the original feature of the interest region, e.g., a microphoneoverlapped with a facial region is inverted into the white beard. As a result,lowering the distortion of the whole image while maintaining the perceptualquality is very challenging. To overcome this trade-off, we propose a simpleyet effective encoder training scheme, coined IntereStyle, which facilitatesencoding by focusing on the interest region. IntereStyle steers the encoder todisentangle the encodings of the interest and uninterest regions. To this end,we filter the information of the uninterest region iteratively to regulate thenegative impact of the uninterest region. We demonstrate that IntereStyleachieves both lower distortion and higher perceptual quality compared to theexisting state-of-the-art encoders. Especially, our model robustly conservesfeatures of the original images, which shows the robust image editing and stylemixing results. We will release our code with the pre-trained model after thereview.

IntereStyle: Encoding an Interest Region for Robust StyleGAN Inversion