gtttttttttttttttttttt/InertialPressureHumanPose

🧍 Human Activity Recognition Dataset with Inertial and Pressure Sensors

📌 Overview

This dataset contains multimodal sensor data collected for human activity recognition using inertial (IMU) and plantar pressure sensors. It was designed to support research in human motion analysis, wearable sensing, smart healthcare, and fall detection.

The dataset includes six types of human activities, with signals recorded from five sensors:

• 3 IM948 inertial sensors (waist, left ankle, right ankle)
• 2 RX-ES42-16P foot pressure sensors (left foot, right foot)

---

📦 Dataset Description

🔧 Sensor Configuration

Sensor Type	Model	Description
Pressure Sensor	RX-ES42-16P	6×4 array per foot, 16 channels × 2 feet = 32 total pressure channels
Inertial Sensor	IM948	Each includes 3-axis accelerometer + 3-axis gyroscope (9 channels total ×3)
Transmission	Bluetooth	All data streamed wirelessly in real time

The RX-ES42-16P uses flexible film array technology to capture high-resolution plantar pressure distribution. The IM948 provides ±16g accelerometer range and ±2000°/s gyroscope range.

🧪 Data Collection

• Participants: Healthy volunteers (balanced genders, 160–180 cm height, 50–80 kg weight)
• Consent: Informed consent was obtained before data collection
• Sensor placement:
• IM948 sensors were attached to the right waist, left ankle, and right ankle
• RX-ES42-16P sensors were inserted under each insole
• Sampling rate: 20 samples persecond
• Each activity was recorded multiple times to ensure reliability

🎯 Activities Covered

Activity Type	Duration per Trial	Notes
Walking	~30 seconds	Repeated across multiple sessions
Standing	~30 seconds	Static posture
Sitting down	~30 seconds	Transition from standing
Stair climbing	~30 seconds	Dynamic behavior
Falling (forward)	~15 seconds	Abnormal event
Falling (sideways)	~15 seconds	Abnormal event

The dataset is imbalanced, with “walking,” “standing,” and “sitting” having the most samples.

---

🧠 Potential Applications

• Human activity recognition
• Fall detection for elderly care
• Gait and posture analysis
• Smart insole systems
• Multimodal time-series fusion models

🙌 Citation

If you use this dataset in your research, please cite:

@misc
{your_dataset_2025,
title={Human Activity Recognition Dataset with Inertial and Pressure Sensors},
author={Yucheng Wang,Xiuyuan Peng,Lingshuo He},
year={2025},
howpublished={\url{https://huggingface.co/datasets/InertialPressureHumanPose}},
}