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

Continual learning is a challenging real-world problem for constructing amature AI system when data are provided in a streaming fashion. Despite recentprogress in continual classification, the researches of continual objectdetection are impeded by the diverse sizes and numbers of objects in eachimage. Different from previous works that tune the whole network for all tasks,in this work, we present a simple and flexible framework for continual objectdetection via pRotOtypical taSk corrElaTion guided gaTing mechAnism (ROSETTA).Concretely, a unified framework is shared by all tasks while task-aware gatesare introduced to automatically select sub-models for specific tasks. In thisway, various knowledge can be successively memorized by storing theircorresponding sub-model weights in this system. To make ROSETTA automaticallydetermine which experience is available and useful, a prototypical taskcorrelation guided Gating Diversity Controller(GDC) is introduced to adaptivelyadjust the diversity of gates for the new task based on class-specificprototypes. GDC module computes class-to-class correlation matrix to depict thecross-task correlation, and hereby activates more exclusive gates for the newtask if a significant domain gap is observed. Comprehensive experiments onCOCO-VOC, KITTI-Kitchen, class-incremental detection on VOC and sequentiallearning of four tasks show that ROSETTA yields state-of-the-art performance onboth task-based and class-based continual object detection.

Continual Object Detection via Prototypical Task Correlation Guided Gating Mechanism