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

3D Single Object Tracking (SOT) is a fundamental task of computer vision,proving essential for applications like autonomous driving. It remainschallenging to localize the target from surroundings due to appearancevariations, distractors, and the high sparsity of point clouds. The spatialinformation indicating objects' spatial adjacency across consecutive frames iscrucial for effective object tracking. However, existing trackers typicallyemploy point-wise representation with irregular formats, leading toinsufficient use of this important spatial knowledge. As a result, thesetrackers usually require elaborate designs and solving multiple subtasks. Inthis paper, we propose BEVTrack, a simple yet effective baseline that performstracking in Bird's-Eye View (BEV). This representation greatly retains spatialinformation owing to its ordered structure and inherently encodes the implicitmotion relations of the target as well as distractors. To achieve accurateregression for targets with diverse attributes (\textit{e.g.}, sizes and motionpatterns), BEVTrack constructs the likelihood function with the learnedunderlying distributions adapted to different targets, rather than making afixed Laplace or Gaussian assumption as in previous works. This providesvaluable priors for tracking and thus further boosts performance. While onlyusing a single regression loss with a plain convolutional architecture,BEVTrack achieves state-of-the-art performance on three large-scale datasets,KITTI, NuScenes, and Waymo Open Dataset while maintaining a high inferencespeed of about 200 FPS. The code will be released athttps://github.com/xmm-prio/BEVTrack.

BEVTrack: A Simple and Strong Baseline for 3D Single Object Tracking in Bird's-Eye View