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 on synthetic data and transferring the resulting properties to theirreal counterparts is an important challenge for reducing costs and increasingsafety in machine learning. In this work, we focus on autoencoder architecturesand aim at learning latent space representations that are invariant toinductive biases caused by the domain shift between simulated and real imagesshowing the same scenario. We train on synthetic images only, presentapproaches to increase generalizability and improve the preservation of thesemantics to real datasets of increasing visual complexity. We show thatpre-trained feature extractors (e.g. VGG) can be sufficient for generalizationon images of lower complexity, but additional improvements are required forvisually more complex scenes. To this end, we demonstrate a new samplingtechnique, which matches semantically important parts of the image, whilerandomizing the other parts, leads to salient feature extraction and aneglection of unimportant parts. This helps the generalization to real data andwe further show that our approach outperforms fine-tuned classification models.

Autoencoder for Synthetic to Real Generalization: From Simple to More Complex Scenes