Core Concepts
Initiating a scalable pipeline for automatic dataset construction and proposing TransAgg model for zero-shot composed image retrieval.
Abstract
The study focuses on Composed Image Retrieval (CIR) to retrieve images using text and image fusion. It introduces a scalable dataset construction pipeline and TransAgg model for zero-shot retrieval, outperforming existing models.
Stats
"Our proposed approach either performs on par with or significantly outperforms the existing state-of-the-art (SOTA) models."
"Our model performs competitively with concurrent work, significantly more efficient."
Quotes
"We propose a retrieval-based pipeline for automatic CIR dataset construction."
"Extensive experiments show that our method performs on par or significantly above the existing state-of-the-art (SOTA) models."