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

Scene flow represents the motion information of each point in the 3D pointclouds. It is a vital downstream method applied to many tasks, such as motionsegmentation and object tracking. However, there are always occlusion pointsbetween two consecutive point clouds, whether from the sparsity data samplingor real-world occlusion. In this paper, we focus on addressing occlusion issuesin scene flow by the semantic self-similarity and motion consistency of themoving objects. We propose a GMA3D module based on the transformer framework,which utilizes local and global semantic similarity to infer the motioninformation of occluded points from the motion information of local and globalnon-occluded points respectively, and then uses an offset aggregator toaggregate them. Our module is the first to apply the transformer-basedarchitecture to gauge the scene flow occlusion problem on point clouds.Experiments show that our GMA3D can solve the occlusion problem in the sceneflow, especially in the real scene. We evaluated the proposed method on theoccluded version of point cloud datasets and get state-of-the-art results onthe real scene KITTI dataset. To testify that GMA3D is still beneficial tonon-occluded scene flow, we also conducted experiments on non-occluded versiondatasets and achieved promising performance on FlyThings3D and KITTI. The codeis available at https://anonymous.4open.science/r/GMA3D-E100.

GMA3D: Local-Global Attention Learning to Estimate Occluded Motions of Scene Flow