1. 四个问题
1. 解决什么问题
步态识别，根据走路识别出 人。
Gait recognition, i.e., recognizing persons from their walking postures, has found versatile applications in security check, health monitoring, and novel human-computer interaction.
2. 用了什么方法解决
- 利用毫米波设备采集数据，做了一个数据集 mmWave gait data set。(基于毫米波 的步态识别 方法具有如下优点：非视觉，又学习了一个新的 模式)
数据集的特点：in which we collect gait of 95 volunteers ’seen’ from two mmWave radars in
two different scenarios, which together lasts about 30 hours.
Compared with traditional camera-based solutions, mmWave based gait recognition bears unique advantages of being still effective under non-line-of-sight scenarios, such as in black, weak light, or blockage conditions. Moreover, they are able to accomplish person identification while preserving privacy. Currently, there are only few works in mmWave gait recognitio
- 提出了一个网络mmGaitNet，5个分支，XYZ（坐标），V（径向速度），S（信号强度），输入是t帧 点云（p x t），5个分支 做特征融合（concat在一起，再用MLP去学（当你不知道怎么实现，就用MLP去学吧）），最后进行行人分类。
3. 效果如何
单人，90%，5个人场景，88%，超过了之前的方法。
compare it with five state-of-the-art algorithms. We find that mmGaitNet is able to achieve 90% accuracy for single-person scenarios, 88% accuracy for five co-existing persons, while
the existing methods achieve less than 66% accuracy for both scenarios.
消融实验：
5个分支：
4. 还存在什么问题
？
2. 论文介绍
3. 参考资料
4. 收获
本文还花了很多篇幅介绍 数据怎么收集，怎么做数据集，但是 我不理解跳过了（怎么从毫米波设备 经过反射形成点云，又怎么计算出点云的5个属性分支？）。
网络是5个分支，输入是t帧 点云（p x t），带时序信息的。
特征融合就是：把特征concat在一起，用MLP去学吧。
又学习到了一个新的模式识别吧。
新的数据集 对于 Gait recognition。
使用毫米波的好处：非视线（such as in black, weak light, or blockage conditions），更好地保护隐私？
分析毫米波为什么有用：带宽大，能够提供细粒度的空间信息。
Moreover, driven by the emerging 5G technologies, gait recognition using the 5G mmWave wireless signal has gained much interest. Compared with the traditional recognition using low-frequency omni-directional Wi-Fi signal (Zou et al. 2018), mmWave gait recognition is promising to achieve higher accuracy, in particular for co-existent multiple persons, because mmWave radios, with 100× bandwidth, can provide much fine-grained spatial resolution.