作者:Alexander HermansLucas BeyerBastian Leibe

作者单位:

发布时间:2017

发布期刊:arxiv

论文全称:In Defense of the Triplet Loss for Person Re-Identification

论文地址:https://arxiv.org/abs/1703.07737

论文代码:https://github.com/VisualComputingInstitute/triplet-reid

地位:

个人理解

一、摘要

In the past few years, the fifield of computer vision has gone through a revolution fueled mainly by the advent of large datasets and the adoption of deep convolutional neural networks for end-to-end learning. The person reidentifification subfifield is no exception to this. Unfortunately, a prevailing belief in the community seems to be that the triplet loss is inferior to using surrogate losses (classififi-cation, verifification) followed by a separate metric learning step. We show that, for models trained from scratch as well as pretrained ones, using a variant of the triplet loss to perform end-to-end deep metric learning outperforms most other published methods by a large margin.

二、研究背景