Enhance Composed Image Retrieval via Multi-level Collaborative Localization and Semantic Activeness Perception

Composed image retrieval (CIR) is an emerging and challenging research task that combines two modalities, a reference image, and a modification text, into one query to retrieve the target image. In online shopping scenarios, the user would use the modification text as feedback to describe the differ...

Ausführliche Beschreibung

Bibliographische Detailangaben

Veröffentlicht in:	IEEE transactions on multimedia 2024-01, Vol.26, p.1-12
Hauptverfasser:	Zhang, Gangjian, Wei, Shikui, Pang, Huaxin, Qiu, Shuang, Zhao, Yao
Format:	Artikel
Sprache:	eng
Schlagworte:	Collaboration Composed Image Retrieval Image enhancement Image retrieval Localization Location awareness Modules Multi-modal Fusion and Embedding Multi-modal Representation Learning Multi-modal Retrieval Retrieval Semantics Task analysis Training Transformers
Online-Zugang:	Volltext bestellen
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!