Similarity Reasoning and Filtration for Image-Text Matching

Image-text matching plays a critical role in bridging the vision and language, great progress has been made by exploiting global alignment between image sentence, or local alignments regions words. However, how to make most of these infer more accurate scores is still underexplored. In this paper, we propose novel Similarity Graph Reasoning Attention Filtration (SGRAF) network for image-text matching. Specifically, vector-based similarity representations are firstly learned characterize comprehensive manner, then (SGR) module relying on one graph convolutional neural introduced relation-aware similarities with both alignments. The (SAF) further developed integrate effectively selectively attending significant representative meanwhile casting aside interferences non-meaningful We demonstrate superiority proposed method achieving state-of-the-art performances Flickr30K MSCOCO datasets, good interpretability SGR SAF extensive qualitative experiments analyses.