WiLib

A toolbox to test deeplearning models on Wi-Fi Human Recognition.

Installation

Try to execute below script on a linux computer with conda and git:

git clone [url]
cd [wilib]
conda create -n wifilib python==3.7
conda activate wifilib
pip install -r requirements.txt

python ./run.py -config [path to config] -debug False

The example configurations are under the 'config' directory.

Model	Paper	Code
THAT	Two-Stream Convolution Augmented Transformer for Human Activity Recognition	https://github.com/windofshadow/THAT
SenseFi	SenseFi: A Library and Benchmark on Deep-Learning-Empowered WiFi Human Sensing	https://github.com/xyanchen/WiFi-CSI-Sensing-Benchmark
ST-GCN (AAAI 2018)	Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition	https://github.com/kennymckormick/pyskl
KAN	KAN: Kolmogorov-Arnold Networks	https://github.com/Blealtan/efficient-kan

Dataset Name	Paper	Data resource	Benchmark
UT-HAR	A Survey on Behavior Recognition Using WiFi Channel State Information	Downloading URL	UTHAR Benchmark
NTU-Fi HAR	SenseFi: A Library and Benchmark on Deep-Learning-Empowered WiFi Human Sensing	Downloading URL	NTU-Fi Benchmark
MM-Fi	MM-Fi: Multi-Modal Non-Intrusive 4D Human Dataset for Versatile Wireless Sensing	Downloading URL	MM-Fi Benchmark