Abstract:The EmoWear dataset provides a bridge to explore Emotion Recognition (ER) via Seismocardiography (SCG), the measurement of small cardio-respiratory induced vibrations on the chest wall through Inertial Measurement Units (IMUs). We recorded Accelerometer (ACC), Gyroscope (GYRO), Electrocardiography (ECG), Blood Volume Pulse (BVP), Respiration (RSP), Electrodermal Activity (EDA), and Skin Temperature (SKT) data from 49 participants who watched validated emotionally stimulating video clips. They self-assessed their emotional valence, arousal, and dominance, as well as extra questions about the video clips. Also, we asked the participants to walk, talk, and drink, so that researchers can detect gait, voice, and swallowing using the same IMU. We demonstrate the effectiveness of emotion stimulation with statistical methods and verify the quality of the collected signals through signal-to-noise ratio and correlation analysis. EmoWear can be used for ER via SCG, ER during gait, multi-modal ER, and the study of IMUs for context-awareness. Targeted contextual information include emotions, gait, voice activity, and drinking, all having the potential to be sensed via a single IMU.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to explore the collection of physiological signals and motion data through wearable devices (especially the chest - worn Inertial Measurement Unit, IMU) in order to achieve Emotion Recognition (ER) and Context Awareness. Specifically, the paper introduces a dataset named EmoWear, which aims to achieve emotion recognition through Seismocardiography (SCG) technology, that is, by using IMU to measure the tiny cardiopulmonary - induced vibrations on the chest. In addition, this dataset also contains the self - assessment data of participants when watching emotional - stimulus video clips, as well as their data during activities such as walking, talking, and drinking water. These data can help researchers detect behaviors such as gait, voice activity, and swallowing. ### Main problems: 1. **Emotion Recognition**: Use physiological signals (such as accelerometer, gyroscope, electrocardiogram (ECG), Blood Volume Pulse (BVP), Respiration (RSP), Electrodermal Activity (EDA) and Skin Temperature (SKT)) collected by IMU through SCG technology to identify the emotional states (emotional valence, arousal and dominance) of participants. 2. **Multi - modal Emotion Recognition**: Combine multiple physiological signals and motion data to improve the accuracy and robustness of emotion recognition. 3. **Context Awareness**: Simultaneously collect multiple context information (such as emotion, gait, voice activity and drinking water) through the same IMU to achieve more comprehensive context awareness. ### Solutions: - **Data Collection**: Collected approximately 70 hours of inertial and physiological recording data from 49 participants. The participants watched 38 verified emotional - stimulus video clips and carried out self - assessments. In addition, they also carried out activities such as walking, reading sentences aloud and drinking water. - **Sensor Configuration**: Used a variety of wearable sensors, including ST SensorTile.box (for accelerometer and gyroscope), Zephyr BioHarness 3 (for ECG, respiration and accelerometer) and Empatica E4 (for BVP, EDA and skin temperature). - **Data Processing and Analysis**: Verified the effectiveness of emotional stimuli through statistical methods, and verified the quality of the collected signals through signal - to - noise ratio and correlation analysis. ### Potential Applications: - **Emotion Recognition**: Achieve more accurate emotion recognition through SCG technology. - **Gait Analysis**: Research the influence of emotional changes during walking. - **Voice Activity Detection**: Use the chest - worn IMU for voice activity detection. - **Swallowing Detection**: Detect drinking water activities through IMU. In conclusion, this paper provides a new research platform through the EmoWear dataset, aiming to achieve multi - aspect physiological and motion data collection through a single chest - worn IMU, thereby promoting research in the fields of emotion recognition and context awareness.

EmoWear: Wearable Physiological and Motion Dataset for Emotion Recognition and Context Awareness

Seismocardiography for Emotion Recognition: A Study on EmoWear with Insights from DEAP

Emotion and Stress Recognition Utilizing Galvanic Skin Response and Wearable Technology: A Real-time Approach for Mental Health Care

WEARS: Wearable Emotion AI with Real-time Sensor data

MDEAW: A Multimodal Dataset for Emotion Analysis through EDA and PPG signals from wireless wearable low-cost off-the-shelf Devices

Temporal Convolutional Network-Enhanced Real-Time Implicit Emotion Recognition with an Innovative Wearable fNIRS-EEG Dual-Modal System

Introducing WESAD, a Multimodal Dataset for Wearable Stress and Affect Detection

Toward Emotion Recognition From Physiological Signals in the Wild: Approaching the Methodological Issues in Real-Life Data Collection

Analyze Spontaneous Gestures for Emotional Stress State Recognition: A Micro-gesture Dataset and Analysis with Deep Learning

Human Emotions Analysis and Recognition Using EEG Signals in Response to 360$^\circ$ Videos

WEMAC: Women and Emotion Multi-modal Affective Computing dataset

An Emotion Recognition System Based on Physiological Signals Obtained by Wearable Sensors

Wearable Emotion Recognition Using Heart Rate Data from a Smart Bracelet

EmpathicSchool: A multimodal dataset for real-time facial expressions and physiological data analysis under different stress conditions

DEAP: A Database for Emotion Analysis ;Using Physiological Signals

ECSMP: A dataset on emotion, cognition, sleep, and multi-model physiological signals

Online Learning for Wearable EEG-Based Emotion Classification

A real-world dataset of group emotion experiences based on physiological data

Emotion Recognition for Everyday Life Using Physiological Signals from Wearables: A Systematic Literature Review

K-EmoCon, a multimodal sensor dataset for continuous emotion recognition in naturalistic conversations