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

Human video motion transfer has a wide range of applications in multimedia,computer vision and graphics. Recently, due to the rapid development ofGenerative Adversarial Networks (GANs), there has been significant progress inthe field. However, almost all existing GAN-based works are prone to addressthe mapping from human motions to video scenes, with scene appearances areencoded individually in the trained models. Therefore, each trained model canonly generate videos with a specific scene appearance, new models are requiredto be trained to generate new appearances. Besides, existing works lack thecapability of appearance control. For example, users have to provide videorecords of wearing new clothes or performing in new backgrounds to enableclothes or background changing in their synthetic videos, which greatly limitsthe application flexibility. In this paper, we propose GAC-GAN, a generalmethod for appearance-controllable human video motion transfer. To enablegeneral-purpose appearance synthesis, we propose to include appearanceinformation in the conditioning inputs. Thus, once trained, our model cangenerate new appearances by altering the input appearance information. Toachieve appearance control, we first obtain the appearance-controllableconditioning inputs and then utilize a two-stage GAC-GAN to generate thecorresponding appearance-controllable outputs, where we utilize an ACGAN lossand a shadow extraction module for output foreground and background appearancecontrol respectively. We further build a solo dance dataset containing a largenumber of dance videos for training and evaluation. Experimental results showthat, our proposed GAC-GAN can not only support appearance-controllable humanvideo motion transfer but also achieve higher video quality than state-of-artmethods.

GAC-GAN: A General Method for Appearance-Controllable Human Video Motion Transfer