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

We present Mix3D, a data augmentation technique for segmenting large-scale 3Dscenes. Since scene context helps reasoning about object semantics, currentworks focus on models with large capacity and receptive fields that can fullycapture the global context of an input 3D scene. However, strong contextualpriors can have detrimental implications like mistaking a pedestrian crossingthe street for a car. In this work, we focus on the importance of balancingglobal scene context and local geometry, with the goal of generalizing beyondthe contextual priors in the training set. In particular, we propose a "mixing"technique which creates new training samples by combining two augmented scenes.By doing so, object instances are implicitly placed into novel out-of-contextenvironments and therefore making it harder for models to rely on scene contextalone, and instead infer semantics from local structure as well. We performdetailed analysis to understand the importance of global context, localstructures and the effect of mixing scenes. In experiments, we show that modelstrained with Mix3D profit from a significant performance boost on indoor(ScanNet, S3DIS) and outdoor datasets (SemanticKITTI). Mix3D can be triviallyused with any existing method, e.g., trained with Mix3D, MinkowskiNetoutperforms all prior state-of-the-art methods by a significant margin on theScanNet test benchmark 78.1 mIoU. Code is available at:https://nekrasov.dev/mix3d/

Mix3D: Out-of-Context Data Augmentation for 3D Scenes