Sampling Matters in Deep Embedding Learning

TL;DR

This paper emphasizes the importance of training example selection in deep embedding learning, proposing distance weighted sampling and demonstrating its effectiveness across multiple datasets.

Contribution

It introduces distance weighted sampling for more effective training example selection and shows that a simple margin-based loss outperforms complex loss functions.

Findings

Distance weighted sampling improves embedding quality.

Margin-based loss outperforms other loss functions.

Achieves state-of-the-art results on multiple datasets.

Abstract

Deep embeddings answer one simple question: How similar are two images? Learning these embeddings is the bedrock of verification, zero-shot learning, and visual search. The most prominent approaches optimize a deep convolutional network with a suitable loss function, such as contrastive loss or triplet loss. While a rich line of work focuses solely on the loss functions, we show in this paper that selecting training examples plays an equally important role. We propose distance weighted sampling, which selects more informative and stable examples than traditional approaches. In addition, we show that a simple margin based loss is sufficient to outperform all other loss functions. We evaluate our approach on the Stanford Online Products, CAR196, and the CUB200-2011 datasets for image retrieval and clustering, and on the LFW dataset for face verification. Our method achieves state-of-the-art performance on all of them.