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

Image-based 3D detection is an indispensable component of the perceptionsystem for autonomous driving. However, it still suffers from the unsatisfyingperformance, one of the main reasons for which is the limited training data.Unfortunately, annotating the objects in the 3D space is extremelytime/resource-consuming, which makes it hard to extend the training setarbitrarily. In this work, we focus on the semi-supervised manner and explorethe feasibility of a cheaper alternative, i.e. pseudo-labeling, to leverage theunlabeled data. For this purpose, we conduct extensive experiments toinvestigate whether the pseudo-labels can provide effective supervision for thebaseline models under varying settings. The experimental results not onlydemonstrate the effectiveness of the pseudo-labeling mechanism for image-based3D detection (e.g. under monocular setting, we achieve 20.23 AP for moderatelevel on the KITTI-3D testing set without bells and whistles, improving thebaseline model by 6.03 AP), but also show several interesting and noteworthyfindings (e.g. the models trained with pseudo-labels perform better than thattrained with ground-truth annotations based on the same training data). We hopethis work can provide insights for the image-based 3D detection community undera semi-supervised setting. The codes, pseudo-labels, and pre-trained modelswill be publicly available.

An Empirical Study of Pseudo-Labeling for Image-based 3D Object Detection