pedropgusmao / operanet

Library to read

OPERAnet

This repository implements functions in Python to read data from OPERAnet, a multimodal activity recognition dataset acquired from radio frequency and vision-based sensors.

Acronyms

• (CSI) WiFi Channel State Information
• (HAR) Human Activity Recognition
• (PWR) Passive WiFi Radar
• (RF) Radio-Frequency
• (UWB) Ultra-Wideband

Install

Dataset pre-processing

The original OPERAnet datasets contain MATLAB String fields that are not compatible with Scipy. To solve this, make sure you run the convert_and_save.m MATLAB script on all directories containing .mat files using the command below. You just have to do this once and it takes a looooong time.

matlab -nodisplay -r "data_dir='/full/path/to/mat/files/directory'; convert_and_save(data_dir); exit"

Install dependencies

Dependencies can be found in the requirements.txt file.

pip install -r requirements.txt

Data Description

WiFi Channel State Information (CSI)

Field	Type	Example
timestamp	datetime	15:07:30.646
activity	List[str]	walk
exp_no	str	exp_002
person_id	str	One
room_no	str	1
tx{1..3}rx{1..3}_sub{1..30}	complex number	11.75 - 1.19j
...	...	...

Passive WiFi Radar (PWR)

Field	Type	Example
exp_no	str	exp_002
timestamp	datetime	15:07:30.646
activity	List[str]	walk
person_id	str	One
room_no	str	1
PWR_ch1	(N,1) float64
PWR_ch2	(N,1) float64
PWR_ch3	(N,1) float64

Ultra-Wideband (UWB)

| Field | Type | TODO

Kinect

Field	Type	Example
exp_no	str	exp_002
timestamp	datetime	15:07:30.646
activity	List[str]	walk
person_id	str	One
room_no	str	1
Kinect1_Markers	(N,3) float64
Kinect2_Markers	(N,3) float64

References

Bocus, M.J., Li, W., Vishwakarma, S. et al. OPERAnet, a multimodal activity recognition dataset acquired from radio frequency and vision-based sensors. Sci Data 9, 474 (2022). https://doi.org/10.1038/s41597-022-01573-2

TODO

• Add description for CSI.
• Convert data types where necessary, i.e. datetime, category, etc.
• WiFi CSI files contain arrays that were stored as strings. Should these be converted back? Should this be done in MATLAB or Python?