papers AI Learner
The Github is limit! Click to go to the new site.

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

2018-07-29
Fartash Faghri, David J. Fleet, Jamie Ryan Kiros, Sanja Fidler

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).

Abstract (translated by Google)
URL

https://arxiv.org/abs/1707.05612

PDF

https://arxiv.org/pdf/1707.05612


Similar Posts

Comments