最近看的行人再识别文章概要

[1] Ejaz Ahmed, Michael Jones, Tim K.Marks.  An Improved Deep Learning Architecture for Person Re-Identification. In CVPR 2015.

　　改变神经网络的架构，提高在benchmark上的performance 使用dnn来做行人再识别(pedestrian re-identification)，引入了两个新的layer—Cross-Input Neighborhood Differences layer、Across Patch Features layer。前者用于比较输入的两幅feature map。并按定义的公式算出两种“differences”，后者用于把两种differences整合起来，变成更高阶的，用于衡量两幅图相似度的特征图。

　　benchmark: CUHK01、CUHK03、VIPeR。

　　取得的实验结果:

　　Our method significantly outperforms the state of the art on both a large data set (CUHK03) and a medium-sized data set (CUHK01), and is resistant to overfitting. We also demonstrate that by initially training on an unrelated large data set before fine-tuning on a small target data set, our network canachieves results comparable to the state of the art even on a small data set (VIPeR).

　　网络拓扑: