A dataset for Wi-Fi-based human-to-human interaction recognition.

Area of Science:

• Computer Science
• Signal Processing
• Human-Computer Interaction

Background:

• Existing Wi-Fi-based human activity datasets primarily focus on single-person activities.
• There is a need for comprehensive datasets to study complex human-to-human interactions using wireless signals.

Purpose of the Study:

• To present a novel dataset for Wi-Fi-based human-to-human interaction recognition.
• To facilitate research in recognizing complex interactions between multiple individuals using wireless signals.

Main Methods:

• Collected 4,800 Wi-Fi signal trials (RSSI and CSI) from 40 pairs performing 12 distinct interactions.
• Utilized a commercial access point (Sagemcom 2704) and an Intel 5300 NIC with the CSI tool.
• Focused on two-person interactions in an indoor environment.

Main Results:

• A dataset comprising 4,800 trials of 12 different human-to-human interactions.
• Data includes Received Signal Strength Indicator (RSSI) and Channel State Information (CSI) values.
• The dataset specifically captures interactions between pairs of subjects.

Conclusions:

• The presented dataset is valuable for advancing Wi-Fi-based human activity recognition, particularly for multi-person interactions.
• It enables the exploration of various machine learning algorithms for recognizing complex human-to-human behaviors.
• This resource supports future research in ubiquitous sensing and human-computer interaction.