Distance learning by mining hard and easy negative samples for person re-identification
Distance learning by mining hard and easy negative samples for person re-identification
复制标题
通过挖掘困难和容易的负样本进行远程学习以进行人员重新识别
DOI：
10.1016/j.patcog.2019.06.007
复制
发表时间：
2019
影响因子：
8
通讯作者：
Cui Xiang
中科院分区：
文献类型：
--
作者：
Zhu Xiaoke;Jing Xiao-Yuan;Zhang Fan;Zhang Xinyu;You Xinge;Cui Xiang
远程学习是一种有效的人再识别技术。在实践中，难负样本通常比易负样本包含更多的判别信息。因此，有必要研究如何在远程学习过程中充分利用不同类型的负样本所传递的判别信息。本文提出了一种基于hard and asynegative samples mining的远程学习（HEND）方法用于人的再识别，该方法通过对hard and asynegative samples设计不同的目标函数来学习距离度量，从而更有效地利用negative samples中包含的判别信息。此外，考虑到不同相机拍摄的图像之间通常存在较大差异，我们进一步提出了一种基于投影的HEND方法，以减少相机之间的差异对重新识别的影响。在7个行人图像数据集上的实验结果验证了所提方法的有效性。
Distance learning is an effective technique for person re-identification. In practice, the hard negative samples usually contain more discriminative information than the easy negative samples. Therefore, it’s necessary to investigate how to make full use of the discriminative information conveyed by different types of negative samples in the distance learning process. In this paper, we propose aHard andEasyNegative samples mining basedDistance learning (HEND) approach for person re-identification, which learns the distance metric by designing different objective functions for hard and easy negative samples, such that the discriminative information contained in negative samples can be exploited more effectively. Moreover, considering that there usually exist large differences between the images captured by different cameras, we further propose a projection-based HEND approach to reduce the influence of between-camera differences to the re-identification. Experimental results on seven pedestrian image datasets demonstrate the effectiveness of the proposed approaches.