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

Transfer learning is a proven technique in 2D computer vision to leverage thelarge amount of data available and achieve high performance with datasetslimited in size due to the cost of acquisition or annotation. In 3D, annotationis known to be a costly task; nevertheless, pre-training methods have onlyrecently been investigated. Due to this cost, unsupervised pre-training hasbeen heavily favored. In this work, we tackle the case of real-time 3D semanticsegmentation of sparse autonomous driving LiDAR scans. Such datasets have beenincreasingly released, but each has a unique label set. We propose here anintermediate-level label set called coarse labels, which can easily be used onany existing and future autonomous driving datasets, thus allowing all the dataavailable to be leveraged at once without any additional manual labeling. Thisway, we have access to a larger dataset, alongside a simple task of semanticsegmentation. With it, we introduce a new pre-training task: coarse labelpre-training, also called COLA. We thoroughly analyze the impact of COLA onvarious datasets and architectures and show that it yields a noticeableperformance improvement, especially when only a small dataset is available forthe finetuning task.

COLA: COarse LAbel pre-training for 3D semantic segmentation of sparse LiDAR datasets