SOTAVerified

VSE++: Improving Visual-Semantic Embeddings with Hard Negatives

2017-07-18Code Available1· sign in to hype

Fartash Faghri, David J. Fleet, Jamie Ryan Kiros, Sanja Fidler

arXivPDF

Code Available — Be the first to reproduce this paper.

Reproduce

Code

Abstract

We present a new technique for learning visual-semantic embeddings for cross-modal retrieval. Inspired by hard negative mining, the use of hard negatives in structured prediction, and ranking loss functions, we introduce a simple change to common loss functions used for multi-modal embeddings. That, combined with fine-tuning and use of augmented data, yields significant gains in retrieval performance. We showcase our approach, VSE++, on MS-COCO and Flickr30K datasets, using ablation studies and comparisons with existing methods. On MS-COCO our approach outperforms state-of-the-art methods by 8.8% in caption retrieval and 11.3% in image retrieval (at R@1).

Tasks

Cross-Modal RetrievalImage RetrievalRetrievalStructured PredictionVisual Reasoning

Benchmark Results

DatasetModelMetricClaimedVerifiedStatus
Flickr30kVSE++ (ResNet)Image-to-text R@152.9—Unverified

Reproductions