cuttercn/gcu_wifisensing

Counting the number of people in an indoor space using Wi-Fi sensing

GCU_WifiSensing

Project name: GCU WiFiSensing (2020.07 ~ 2021.06)
Team members: 이재윤(Jaeyoon Lee), 정은서(Eunseo Jeong), 여찬영(Chanyeong Yeo)

System architecture

In our project, Fair MOT and Linux 802.11n CSI tool library were used. Each library was used when recognizing and tracking objects from the video data received from the initial camera and when collecting and quantifying Wi-Fi CSI data from the AP.

CSI

• We use the 802.11n CSI Tool for Wi-Fi Sensing. So Reference this link and install 802.11n CSI Tool
• Install Matlab program, and Downlad Our CSI file(Reference for Realtime-processing-for-csitool, linux-80211n-csitool-supplementary) this file read csi data and output csi matrix
• Install success next step

1. our File:

    git clone https://github.com/jyoonlee/GCU_WifiSensing.git
   

2. In matlab:

    cd CSI
    cd matlab
    run read_bf_socket using Matlab
   

3. In terminal:

   sudo stop network-manager
   sudo modprobe -r iwlwifi mac80211
   sudo modprobe iwlwifi connector_log=0x1
   

4. In other terminal(connect Wifi):

    iw dev
    sudo ip link show wlan0
    sudo ip link set wlan0 up
    iw wlan0 link
    sudo iw dev wlan0 connect [WiFi name]
    iw wlan0 link
    sudo dhclient wlan0
   

5. connet csi:

    cd CSI
    cd linux-80211n-csitool-supplementary-master/netlink
    gcc log_to_server.c -o log_to_server
    sudo ./log_to_server 127.0.0.1 8090
   

6. ping test

    ping -i 0.2 192.168.1.1
   

FairMOT

• We used FairMOT model and changed the code according to us.

• We referred to this link for FairMOT code.

• installation

    conda create -n FairMOT
    conda activate FairMOT
    conda install pytorch==1.7.0 torchvision==0.8.0 cudatoolkit=10.2 -c pytorch
    cd ${FAIRMOT_ROOT}
    pip install -r requirements.txt
  

• baseline model pretrain model fairmot_dla34.pth Google Reference this link Model save structure

      ${FAIRMOT_ROOT}
       └——————models
               └——————fairmot_dla34.pth
               └——————...
       └——————src
               └——————...
       └——————video
               └——————video
       └——————demo
               └——————result.txt
               └——————frame image
               └——————output.mp4
               └——————...
  

• Use FairMOT Object Detection

    cd ${FAIRMOT_ROOT(Object Detection)}
    cd src
    python demo.py mot --load_model ../models/fairmot_dla34.pth --conf_thres 0.4
  

• result.txt output

    YY-MM-DD hh:mm:ss       object label
  

• Use FairMOT position

    cd ${FAIRMOT_ROOT(Position)}
    cd src
    python demo.py mot --load_model ../models/fairmot_dla34.pth --conf_thres 0.4
  

• result.txt output

    frame_number    Object id       position x      position y
  

Experiment

• Use TrainModel.py

    python TrainModel.py
  

• input

1. CSI Data
2. Object Detection Result.txt

• Progress

1. synchronize csi label and Result.txt
2. Use Randomforest

Result

CSI Demo Video

MOT Demo Video

Experiment