WIDAR3.0: WiFi-based Activity Recognition Dataset

This repository hosts the data and code for paper "Zero-Effort Cross-Domain Gesture Recognition with Wi-Fi" published in ACM MobiSys 2019.This dataset includes the Channel State Information (CSI) data collected from commodity Wi-Fi devices for gestures and Body-coordinate Velocity Profile (BVP) data calculated by the algorithms from the above mentioned paper.Also included in this repository is the data for paper "GaitID: Robust WiFi Based Gait Recognition" published in Springer WASA 2020.Please stay tuned for further updates.Due to technical issues, please access our dataset homepage for download links: http://tns.thss.tsinghua.edu.cn/widar3.0/本仓库收录了发表于ACM MobiSys 2019会议的论文《Wi-Fi零开销跨域手势识别》（Zero-Effort Cross-Domain Gesture Recognition with Wi-Fi）对应的数据集与代码。该数据集包含从商用Wi-Fi设备采集的手势相关信道状态信息（Channel State Information, CSI）数据，以及由上述论文提出的算法计算得到的身体坐标速度剖面（Body-coordinate Velocity Profile, BVP）数据。本仓库同时收录了发表于Springer WASA 2020的论文《GaitID：基于Wi-Fi的鲁棒步态识别》（GaitID: Robust WiFi Based Gait Recognition）对应的数据集。敬请期待后续更新。由于技术问题，请访问我们的数据集主页获取下载链接：http://tns.thss.tsinghua.edu.cn/widar3.0/

搜集汇总

背景与挑战

背景概述

WIDAR3.0是一个大规模、多域的WiFi活动识别数据集，专为手势识别设计，包含25.8万个手势实例，总时长8620分钟，覆盖75个域。数据集采集自商用WiFi网卡的RSSI和CSI射频信号，并提供了DFS和BVP等高级特征，适用于射频信号处理与机器学习研究。