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

Current LiDAR-based 3D object detectors for autonomous driving are almostentirely trained on human-annotated data collected in specific geographicaldomains with specific sensor setups, making it difficult to adapt to adifferent domain. MODEST is the first work to train 3D object detectors withoutany labels. Our work, HyperMODEST, proposes a universal method implemented ontop of MODEST that can largely accelerate the self-training process and doesnot require tuning on a specific dataset. We filter intermediate pseudo-labelsused for data augmentation with low confidence scores. On the nuScenes dataset,we observe a significant improvement of 1.6% in AP BEV in 0-80m range atIoU=0.25 and an improvement of 1.7% in AP BEV in 0-80m range at IoU=0.5 whileonly using one-fifth of the training time in the original approach by MODEST.On the Lyft dataset, we also observe an improvement over the baseline duringthe first round of iterative self-training. We explore the trade-off betweenhigh precision and high recall in the early stage of the self-training processby comparing our proposed method with two other score filtering methods:confidence score filtering for pseudo-labels with and without static labelretention. The code and models of this work are available athttps://github.com/TRAILab/HyperMODEST

HyperMODEST: Self-Supervised 3D Object Detection with Confidence Score Filtering