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

Deep learning-based approaches have shown remarkable performance in the 3Dobject detection task. However, they suffer from a catastrophic performancedrop on the originally trained classes when incrementally learning new classeswithout revisiting the old data. This "catastrophic forgetting" phenomenonimpedes the deployment of 3D object detection approaches in real-worldscenarios, where continuous learning systems are needed. In this paper, westudy the unexplored yet important class-incremental 3D object detectionproblem and present the first solution - SDCoT, a novel static-dynamicco-teaching method. Our SDCoT alleviates the catastrophic forgetting of oldclasses via a static teacher, which provides pseudo annotations for old classesin the new samples and regularizes the current model by extracting previousknowledge with a distillation loss. At the same time, SDCoT consistently learnsthe underlying knowledge from new data via a dynamic teacher. We conductextensive experiments on two benchmark datasets and demonstrate the superiorperformance of our SDCoT over baseline approaches in several incrementallearning scenarios.

Static-Dynamic Co-Teaching for Class-Incremental 3D Object Detection