Directional Statistics-based Deep Metric Learning for Image Classification and Retrieval

TL;DR

This paper introduces a novel deep metric learning approach using von Mises-Fisher distribution and cosine similarity, achieving state-of-the-art results in image classification and retrieval with a simpler training process.

Contribution

It proposes a new loss function based on von Mises-Fisher distribution and a global structure learning algorithm for hyper-spherical embedding, improving over Euclidean-based methods.

Findings

Achieves state-of-the-art performance on standard datasets.

Outperforms softmax loss with fewer convolutional layers.

Simplifies training procedure for deep metric learning.

Abstract

Deep distance metric learning (DDML), which is proposed to learn image similarity metrics in an end-to-end manner based on the convolution neural network, has achieved encouraging results in many computer vision tasks.$L2$-normalization in the embedding space has been used to improve the performance of several DDML methods. However, the commonly used Euclidean distance is no longer an accurate metric for $L2$-normalized embedding space, i.e., a hyper-sphere. Another challenge of current DDML methods is that their loss functions are usually based on rigid data formats, such as the triplet tuple. Thus, an extra process is needed to prepare data in specific formats. In addition, their losses are obtained from a limited number of samples, which leads to a lack of the global view of the embedding space. In this paper, we replace the Euclidean distance with the cosine similarity to better utilize the $L2$-normalization, which is able to attenuate the curse of dimensionality. More specifically, a novel loss function based on the von Mises-Fisher distribution is proposed to learn a compact hyper-spherical embedding space. Moreover, a new efficient learning algorithm is developed to better capture the global structure of the embedding space. Experiments for both classification and retrieval tasks on several standard datasets show that our method achieves state-of-the-art performance with a simpler training procedure. Furthermore, we demonstrate that, even with a small number of convolutional layers, our model can still obtain significantly better classification performance than the widely used softmax loss.