.png){kind=icon}
.png){kind=icon}
Return
Scene graph optimization using hyperedge-based relationship extraction
.png){kind=icon}
PRE
AI
DOI：10.1016/j.asoc.2026.114907.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Scene graph optimization via hyperedge-based relationship extraction is proposed. • Association-rule mining filters redundant relations to enhance graph efficiency. • Hypergraph partitioning simplifies structures while preserving semantic context. • Relation mining improves recall by assigning weights to key object detections. • The proposed method achieves superior BLEU and ROUGE scores on COCO datasets.
Keywords:
scene graph optimization
hyperedge-based relationship extraction
association-rule mining
hypergraph partitioning
relation mining