A Metric Learning Reality Check

Abstract

Deep metric learning papers from the past four years have consistently claimed great advances in accuracy, often more than doubling the performance of decade-old methods. In this paper, we take a closer look at the field to see if this is actually true. We find flaws in the experimental methodology of numerous metric learning papers, and show that the actual improvements over time have been marginal at best.

Visit https://www.github.com/KevinMusgrave/powerful-benchmarker for supplementary material, including the source code, configuration files, log files, and interactive bayesian optimization plots