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

We contribute the Habitat Synthetic Scene Dataset, a dataset of 211high-quality 3D scenes, and use it to test navigation agent generalization torealistic 3D environments. Our dataset represents real interiors and contains adiverse set of 18,656 models of real-world objects. We investigate the impactof synthetic 3D scene dataset scale and realism on the task of trainingembodied agents to find and navigate to objects (ObjectGoal navigation). Bycomparing to synthetic 3D scene datasets from prior work, we find that scalehelps in generalization, but the benefits quickly saturate, making visualfidelity and correlation to real-world scenes more important. Our experimentsshow that agents trained on our smaller-scale dataset can match or outperformagents trained on much larger datasets. Surprisingly, we observe that agentstrained on just 122 scenes from our dataset outperform agents trained on 10,000scenes from the ProcTHOR-10K dataset in terms of zero-shot generalization inreal-world scanned environments.

Habitat Synthetic Scenes Dataset (HSSD-200): An Analysis of 3D Scene Scale and Realism Tradeoffs for ObjectGoal Navigation