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

Scene Graph Generation (SGG) offers a structured representation critical inmany computer vision applications. Traditional SGG approaches, however, arelimited by a closed-set assumption, restricting their ability to recognize onlypredefined object and relation categories. To overcome this, we categorize SGGscenarios into four distinct settings based on the node and edge: Closed-setSGG, Open Vocabulary (object) Detection-based SGG (OvD-SGG), Open VocabularyRelation-based SGG (OvR-SGG), and Open Vocabulary Detection + Relation-basedSGG (OvD+R-SGG). While object-centric open vocabulary SGG has been studiedrecently, the more challenging problem of relation-involved open-vocabulary SGGremains relatively unexplored. To fill this gap, we propose a unified frameworknamed OvSGTR towards fully open vocabulary SGG from a holistic view. Theproposed framework is an end-toend transformer architecture, which learns avisual-concept alignment for both nodes and edges, enabling the model torecognize unseen categories. For the more challenging settings ofrelation-involved open vocabulary SGG, the proposed approach integratesrelation-aware pre-training utilizing image-caption data and retainsvisual-concept alignment through knowledge distillation. Comprehensiveexperimental results on the Visual Genome benchmark demonstrate theeffectiveness and superiority of the proposed framework.

Expanding Scene Graph Boundaries: Fully Open-vocabulary Scene Graph Generation via Visual-Concept Alignment and Retention