Deep Multimodal Image-Text Embeddings for Automatic Cross-Media Retrieval

This paper considers the task of matching images and sentences by learning a visual-textual embedding space for cross-modal retrieval. Finding such a space is a challenging task since the features and representations of text and image are not comparable. In this work, we introduce an end-to-end deep multimodal convolutional-recurrent network for learning both vision and language representations simultaneously to infer image-text similarity. The model learns which pairs are a match (positive) and which ones are a mismatch (negative) using a hinge-based triplet ranking. To learn about the joint representations, we leverage our newly extracted collection of tweets from Twitter. The main characteristic of our dataset is that the images and tweets are not standardized the same as the benchmarks. Furthermore, there can be a higher semantic correlation between the pictures and tweets contrary to benchmarks in which the descriptions are well-organized. Experimental results on MS-COCO benchmark dataset show that our model outperforms certain methods presented previously and has competitive performance compared to the state-of-the-art. The code and dataset have been made available publicly.

Hadi Abdi Khojasteh, Ebrahim Ansari, Parvin Razzaghi, Akbar Karimi (4) __INSTITUTION_4__ Institute for Advanced Studies in Basic Sciences __INSTITUTION_5__, Zanjan, Iran, (2) Faculty of Mathematics, Physics, Institute of Formal, Applied Linguistics, Charles University, Czechia, (3) Institute for Research in Fundamental Sciences __INSTITUTION_8__, Tehran, Iran, (4) IMP Lab, Department of Engineering, Architecture, University of Parma, Parma, Italy) · 2020