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 single object tracking is a key task in 3D computer vision. However, thesparsity of point clouds makes it difficult to compute the similarity andlocate the object, posing big challenges to the 3D tracker. Previous workstried to solve the problem and improved the tracking performance in some commonscenarios, but they usually failed in some extreme sparse scenarios, such asfor tracking objects at long distances or partially occluded. To address theabove problems, in this letter, we propose a sparse-to-dense andtransformer-based framework for 3D single object tracking. First, we transformthe 3D sparse points into 3D pillars and then compress them into 2D BEVfeatures to have a dense representation. Then, we propose an attention-basedencoder to achieve global similarity computation between template and searchbranches, which could alleviate the influence of sparsity. Meanwhile, theencoder applies the attention on multi-scale features to compensate for thelack of information caused by the sparsity of point cloud and the single scaleof features. Finally, we use set-prediction to track the object through atwo-stage decoder which also utilizes attention. Extensive experiments showthat our method achieves very promising results on the KITTI and NuScenesdatasets.

Exploiting More Information in Sparse Point Cloud for 3D Single Object Tracking