DeepEMD: Few-Shot Image Classification With Differentiable Earth Mover’s Distance and Structured Classifiers

TLDR

Earth Mover’s Distance computes optimal matching flows between structural elements, providing a distance metric for image classification. The study proposes a novel few‑shot classification approach that uses optimal matching of image regions via Earth Mover’s Distance. The method employs Earth Mover’s Distance as a differentiable layer, augmented with a cross‑reference weighting scheme and a structured fully connected classifier to learn dense image representations for few‑shot tasks. The approach achieves state‑of‑the‑art results on miniImageNet, tieredImageNet, FC100, and CUB few‑shot benchmarks.

Abstract

In this paper, we address the few-shot classification task from a new perspective of optimal matching between image regions. We adopt the Earth Mover's Distance (EMD) as a metric to compute a structural distance between dense image representations to determine image relevance. The EMD generates the optimal matching flows between structural elements that have the minimum matching cost, which is used to represent the image distance for classification. To generate the important weights of elements in the EMD formulation, we design a cross-reference mechanism, which can effectively minimize the impact caused by the cluttered background and large intra-class appearance variations. To handle k-shot classification, we propose to learn a structured fully connected layer that can directly classify dense image representations with the EMD. Based on the implicit function theorem, the EMD can be inserted as a layer into the network for end-to-end training. We conduct comprehensive experiments to validate our algorithm and we set new state-of-the-art performance on four popular few-shot classification benchmarks, namely miniImageNet, tieredImageNet, Fewshot-CIFAR100 (FC100) and Caltech-UCSD Birds-200-2011 (CUB).