Embedded Real-Time Fall Detection with Deep Learning on Wearable Devices

Torti, Emanuele; Fontanella, Alessandro; Musci, Mirto; Blago, Nicola; Pau, Danilo; Leporati, Francesco; Piastra, Marco

doi:10.1109/dsd.2018.00075

Cited by 55 publications

(58 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…e board is also equipped with 1 MB of flash memory and 128 KB of SRAM. is device has been successfully employed in human activity recognition [21] and fall detection [4]. In the above perspective, the fall detection system presented in [4] performs data classification through deep learning methods elaborated on the device.…”

Section: State Of the Artmentioning

confidence: 99%

“…is device has been successfully employed in human activity recognition [21] and fall detection [4]. In the above perspective, the fall detection system presented in [4] performs data classification through deep learning methods elaborated on the device. is means that it outperforms the other devices in terms of accuracy because it adopts deep learning methods and reduces power consumption since the elaboration is performed on board, without the need for continuous data transfers.…”

Section: State Of the Artmentioning

confidence: 99%

“…However, this system can be further optimized both from the point of view of computational complexity and power consumption. In this paper, we present a significant improvement of our previous work described in [4]; in particular, the system has been enriched with a pre-elaboration step which prevents the MCU from classifying data that does not contain relevant information. Moreover, a communication protocol based on the BLE standard has been designed in order to 2 Scientific Programming minimize the communications between the SensorTile board and a remote host.…”

Section: State Of the Artmentioning

confidence: 99%

“…Moreover, the part of the complete TensorFlow framework which is implemented is at present insufficient to implement LSTMs. To the best of the authors' knowledge, RNNs have been successfully deployed to an ARM Cortex M4 MCU only in our previous work [4], where we described a runtime inference module based on an RNN network. In this work, the communication module plus other energy-saving provisions were not discussed and, as we will see here, such improvements can be significantly beneficial to the whole embedded module.…”

Section: Fall Detection With Deep Learning For Embeddedmentioning

confidence: 99%

“…In this paper, we describe an embedded framework based on edge computing and on machine learning for fall detection on-board wearable devices. In particular, we extend our system proposed in [4] with a novel strategy to improve the overall system performance together with the design of a suitable Bluetooth Low Energy (BLE) protocol for an effective minimization of data transmission.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Deep Recurrent Neural Networks for Edge Monitoring of Personal Risk and Warning Situations

Torti

Musci

Guareschi

et al. 2019

Scientific Programming

Self Cite

View full text Add to dashboard Cite

Accidental falls are the main cause of fatal and nonfatal injuries, which typically lead to hospital admissions among elderly people. A wearable system capable of detecting unintentional falls and sending remote notifications will clearly improve the quality of the life of such subjects and also helps to reduce public health costs. In this paper, we describe an edge computing wearable system based on deep learning techniques. In particular, we give special attention to the description of the classification and communication modules, which have been developed by keeping in mind the limits in terms of computational power, memory occupancy, and power consumption of the designed wearable device. e system thus developed is capable of classifying 3Daccelerometer signals in real-time and to issue remote alerts while keeping power consumption low and improving on the present state-of-the-art solutions in the literature.

show abstract

Section: State Of the Artmentioning

confidence: 99%

Section: State Of the Artmentioning

confidence: 99%

Section: State Of the Artmentioning

confidence: 99%

Section: Fall Detection With Deep Learning For Embeddedmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Deep Recurrent Neural Networks for Edge Monitoring of Personal Risk and Warning Situations

Torti

Musci

Guareschi

et al. 2019

Scientific Programming

Self Cite

View full text Add to dashboard Cite

show abstract

An Automated Fall Detection System Using Recurrent Neural Networks

Luna-Perejón

Civit-Masot

Amaya-Rodriguez

et al. 2019

Artificial Intelligence in Medicine

View full text Add to dashboard Cite

Falls are the most common cause of fatal injuries in elderly people, causing even death if there is no immediate assistance. Fall detection systems can be used to alert and request help when this type of accident happens. Certain types of these systems include wearable devices that analyze bio-medical signals from the person carrying it in real time. In this way, Deep Learning algorithms could automate and improve the detection of unintentional falls by analyzing these signals. These algorithms have proven to achieve high effectiveness with competitive performances in many classification problems. This work aims to study 16 Recurrent Neural Networks architectures (using Long Short-Term Memory and Gated Recurrent Units) for falls detection based on accelerometer data, reducing computational requirements of previous research. The architectures have been tested on a labeled version of the publicly available SisFall dataset, achieving a mean F1-score above 0.73 and improving state-of-the-art solutions in terms of network complexity.

show abstract