W WiSenseHub

AGENT-READY WIFI SENSING DATA

WiFi sensing data, ready to use.

Find a dataset, ask an agent to prepare it, or follow the guided workflow yourself. WiSenseHub handles the format differences and checks the result.

Prepare with an agent Browse datasets
Person walking through a furnished room between two WiFi devices
Ask in plain language “Prepare this WiFi sensing activity dataset.”
DownloadedOrganizedChecked
—dataset entries
—task categories
—hardware settings
—direct-access releases

USE WITH YOUR AGENT

Install once. Ask in plain language.

The open WiSenseHub Agent Skill works with Codex, Claude Code, Cursor, and other Agent Skills-compatible tools.

ONE LINE TO START

Paste this into your coding agent:

Install the WiSenseHub skill by running npx skills add wisensehub/wisensehub.github.io --skill wisensehub -g, then help me prepare WiFi sensing data.

Once installed, describe the dataset and result you want.

TRY ASKING
  • “Download the CSI-Bench sample, standardize it, and show me what changed.”
  • “Prepare an activity-recognition split and summarize the labels.”
  • “Inspect my CSI files, fix format problems, and visualize one sample.”

NEW TO WIFI SENSING?

Start with applications and basic concepts

A gentler page explains CSI, downstream use cases, clinical and smart-home scenarios, and learning resources before the command line.

For clinicians, educators, smart-home researchers, and students

Learn what channel state information means, why motion changes WiFi signals, where WiFi sensing can be useful, and how WiSenseHub turns collected wireless data into inspectable research files.

Open beginner page

CATALOG

Find data by research question

Every entry distinguishes discovery, download access, and redistribution rights.

WHAT WISENSEHUB DOES

Two steps: schema, then task preprocess

Native files stay intact. Fixed-size model views are derived and audited.

01 · Schema

Standardize structure

Convert heterogeneous releases into one NPZ + JSON contract with explicit axes, masks, labels, and provenance.

[T, S, Tx, Rx]+metadata.json+valid_mask
02 · Task view

Preprocess per task profile

Resample, crop, or pad time so clips for the same task share rate, duration, and layout—without inventing absolute calibration.

profile Hz·duration / T·layout

Read the shared schema, preprocessing, mask, and Python guide →

CITE THIS WORK

Use the software citation and repository URL

If WiSenseHub supports your research, please cite the software and the original dataset releases.

WiSenseHub: A Task-Oriented WiFi Sensing Data Hub.

Repository: https://github.com/wisensehub/wisensehub.github.io