| 研究課題/領域番号 |
18K11408
|
|---|
| 研究機関 | 法政大学 |
|---|
研究代表者 |
Jianhua Ma 法政大学, 情報科学部, 教授 (70295426)
|
|---|
| 研究分担者 |
Huang Runhe 法政大学, 情報科学部, 教授 (00254102)
|
|---|
| 研究期間 (年度) |
2018-04-01 – 2023-03-31
|
| キーワード | Wearable / Recognition / Human / Dog / Activity / Emotion |
|---|
| 研究実績の概要 |
Our research in 2021 was focused mainly on the following aspects. First, we incorporated other devices with various kinds of sensor in our research. These devices are (1) M5StickC with embedded sensors of acceleration and gyroscope as well as extended sensors of force pressure and air pressure; (2) Go Direct174; Respiration Belt to sense the breath data; (3) Polar H10 to get high accurate heart rate variability (HRV). Second, based on these sensors we have done a variety of on anomaly detection, activity recognition such as washing hand and emotion recognition such as fatigue estimation using deep learning algorithms of CNN and LSTM. Third, we also used a millimeter Wave radar to measure vital signs in a contactless manner together with using wearable devices as the reference. Our research results have been published in the IEEE international conference CyberSciTech and journals.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
2: おおむね順調に進展している
理由
The research in 2021 fiscal year was carried out basically as we planned. One of the basic characteristics of this proposed Wear-I research is to apply various wearable sensors in different kinds of wearable applications. In 2021, we have used a variety of sensors including M5StickC, WitMotion, SEEEDRRI-110990038, SeeedGrove101020068, ENV II Hat, MUSE2, Go Direct174; Respiration Belt and Polar H10 to obtain different types of data. These data are mainly used for recognition of activity and emotion. In particular, we have done recognition and change detection of nine handwashing activities using wristed wearables and archived the high recognition accuracy over 99%. As for emotion recognition, we have done research on positive emotion detection, fear level estimation and the recognition of physical and psychological fatigues. Our research has shown that the LSTM is slightly better than CNN in making the recognition. We have further extended the wearable application from human to dogs.
|
| 今後の研究の推進方策 |
The research in 2022 will be focused mainly on applying multiple devices for monitoring daily life covering both human and dogs. We are going to conduct the following research. We shall research on monitoring sleep, which is one of the common and important human activities. Both wearable devices and non-contact devices such as radar will be used together for sleep monitoring. Due to the pandemic of COVID-19, the online learning and working have become more popular. Therefore, we plan to study monitoring both physical and emotional states in such online learning/working. We are going to focus on emotion recognition of stress level, concentration degree, arousal extent and valence states based on physiological data such as brainwave, heart rate and breath pattern. We shall continue to study the dog monitoring including dog vital signs and activities. The main purpose of these studies is to make relatively practical applications using various devices by deep leaning algorithms.
|
| 次年度使用額が生じた理由 |
Due to the pandemic of COVID-19, it was difficult to conduct all onsite experiments involving with sufficient subjects. This situation is expected to get better in 2022, and it would be possible to conduct various experiments.
|
