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

3D object detection has seen quick progress thanks to advances in deeplearning on point clouds. A few recent works have even shown state-of-the-artperformance with just point clouds input (e.g. VoteNet). However, point clouddata have inherent limitations. They are sparse, lack color information andoften suffer from sensor noise. Images, on the other hand, have high resolutionand rich texture. Thus they can complement the 3D geometry provided by pointclouds. Yet how to effectively use image information to assist point cloudbased detection is still an open question. In this work, we build on top ofVoteNet and propose a 3D detection architecture called ImVoteNet specializedfor RGB-D scenes. ImVoteNet is based on fusing 2D votes in images and 3D votesin point clouds. Compared to prior work on multi-modal detection, we explicitlyextract both geometric and semantic features from the 2D images. We leveragecamera parameters to lift these features to 3D. To improve the synergy of 2D-3Dfeature fusion, we also propose a multi-tower training scheme. We validate ourmodel on the challenging SUN RGB-D dataset, advancing state-of-the-art resultsby 5.7 mAP. We also provide rich ablation studies to analyze the contributionof each design choice.

ImVoteNet: Boosting 3D Object Detection in Point Clouds with Image Votes