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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Recent advancements in deep learning opened new opportunities for learning ahigh-quality 3D model from a single 2D image given sufficient training onlarge-scale data sets. However, the significant imbalance between availableamount of images and 3D models, and the limited availability of labeled 2Dimage data (i.e. manually annotated pairs between images and theircorresponding 3D models), severely impacts the training of most supervised deeplearning methods in practice. In this paper, driven by a novel design ofadversarial networks, we have developed an unsupervised learning paradigm toreconstruct 3D models from a single 2D image, which is free of manuallyannotated pairwise input image and its associated 3D model. Particularly, theparadigm begins with training an adaption network via autoencoder withadversarial loss, which embeds unpaired 2D synthesized image domain with realworld image domain to a shared latent vector space. Then, we jointly train a 3Ddeconvolutional network to transform the latent vector space to the 3D objectspace together with the embedding process. Our experiments verify our network'srobust and superior performance in handling 3D volumetric object generationfrom a single 2D image.

Unsupervised 3D Reconstruction from a Single Image via Adversarial Learning