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

The remarkable breakthroughs in point cloud representation learning haveboosted their usage in real-world applications such as self-driving cars andvirtual reality. However, these applications usually have an urgent requirementfor not only accurate but also efficient 3D object detection. Recently,knowledge distillation has been proposed as an effective model compressiontechnique, which transfers the knowledge from an over-parameterized teacher toa lightweight student and achieves consistent effectiveness in 2D vision.However, due to point clouds' sparsity and irregularity, directly applyingprevious image-based knowledge distillation methods to point cloud detectorsusually leads to unsatisfactory performance. To fill the gap, this paperproposes PointDistiller, a structured knowledge distillation framework forpoint clouds-based 3D detection. Concretely, PointDistiller includes localdistillation which extracts and distills the local geometric structure of pointclouds with dynamic graph convolution and reweighted learning strategy, whichhighlights student learning on the crucial points or voxels to improveknowledge distillation efficiency. Extensive experiments on both voxels-basedand raw points-based detectors have demonstrated the effectiveness of ourmethod over seven previous knowledge distillation methods. For instance, our 4Xcompressed PointPillars student achieves 2.8 and 3.4 mAP improvements on BEVand 3D object detection, outperforming its teacher by 0.9 and 1.8 mAP,respectively. Codes have been released athttps://github.com/RunpeiDong/PointDistiller.

PointDistiller: Structured Knowledge Distillation Towards Efficient and Compact 3D Detection