Sampling Matters in Deep Embedding Learning
This addresses the challenge of improving similarity-based tasks like verification and search for computer vision applications, though it is incremental by focusing on sampling rather than introducing a new paradigm.
The paper tackles the problem of selecting training examples in deep embedding learning, showing that it is as important as the loss function, and achieves state-of-the-art performance on multiple datasets for image retrieval, clustering, and face verification.
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.