A Metric Learning Reality Check

Code

• github.com/KevinMusgrave/powerful-benchmarker
  OfficialIn paperpytorch★ 439
• github.com/KevinMusgrave/pytorch-metric-learning
  pytorch★ 6,311
• github.com/bnu-wangxun/Deep_Metric
  pytorch★ 779
• github.com/alpoler/mohan2020
  pytorch★ 4

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.

Tasks

Reproductions