Automatic Synchronization of Wearable Sensors and Video-Cameras for Ground Truth Annotation -- A Practical Approach

Author(s):  
Thomas Plotz ◽  
Chen Chen ◽  
Nils Y. Hammerla ◽  
Gregory D. Abowd
2020 ◽  
Vol 64 (5) ◽  
pp. 50411-1-50411-8
Author(s):  
Hoda Aghaei ◽  
Brian Funt

Abstract For research in the field of illumination estimation and color constancy, there is a need for ground-truth measurement of the illumination color at many locations within multi-illuminant scenes. A practical approach to obtaining such ground-truth illumination data is presented here. The proposed method involves using a drone to carry a gray ball of known percent surface spectral reflectance throughout a scene while photographing it frequently during the flight using a calibrated camera. The captured images are then post-processed. In the post-processing step, machine vision techniques are used to detect the gray ball within each frame. The camera RGB of light reflected from the gray ball provides a measure of the illumination color at that location. In total, the dataset contains 30 scenes with 100 illumination measurements on average per scene. The dataset is available for download free of charge.


Sensors ◽  
2020 ◽  
Vol 20 (16) ◽  
pp. 4474 ◽  
Author(s):  
Tal Reches ◽  
Moria Dagan ◽  
Talia Herman ◽  
Eran Gazit ◽  
Natalia A. Gouskova ◽  
...  

Freezing of gait (FOG) is a debilitating motor phenomenon that is common among individuals with advanced Parkinson’s disease. Objective and sensitive measures are needed to better quantify FOG. The present work addresses this need by leveraging wearable devices and machine-learning methods to develop and evaluate automated detection of FOG and quantification of its severity. Seventy-one subjects with FOG completed a FOG-provoking test while wearing three wearable sensors (lower back and each ankle). Subjects were videotaped before (OFF state) and after (ON state) they took their antiparkinsonian medications. Annotations of the videos provided the “ground-truth” for FOG detection. A leave-one-patient-out validation process with a training set of 57 subjects resulted in 84.1% sensitivity, 83.4% specificity, and 85.0% accuracy for FOG detection. Similar results were seen in an independent test set (data from 14 other subjects). Two derived outcomes, percent time frozen and number of FOG episodes, were associated with self-report of FOG. Bother derived-metrics were higher in the OFF state than in the ON state and in the most challenging level of the FOG-provoking test, compared to the least challenging level. These results suggest that this automated machine-learning approach can objectively assess FOG and that its outcomes are responsive to therapeutic interventions.


Author(s):  
Idongesit Ekerete ◽  
Chris Nugent ◽  
James McLaughlin

This paper proposes the localisation of room occupants in home environments using Unobtrusive Sensing Solutions (USSs). The ability to localise room occupants in home environments can help in the objective monitoring of sedentary behaviour. While wearable sensors can provide tangible information on health and wellness, they have battery life issues and the inability to perform prolonged monitoring. This work uses heterogeneous USSs in the form of an Infrared Thermopile Array (ITA-64) thermal sensor and a Multi-Chirp Frequency Modulated Continuous Wave Mono-pulse (MC-FMCW-M) Radar sensor to monitor room occupants. Digital filters and background subtraction algorithms were used to process the thermal images gleaned from the ITA-64 thermal sensors. The MC-FMCW-M Radar sensor used multi-chirp and Doppler shift principles to estimate the exact location of the targeted room occupants. The estimated distances from the Radar Sensor were compared with ground truth values. Experimental results demonstrated the ability to identify thermal blobs of occupants present in the room at any particular time. Data analyses indicated no significant difference (p = 0.975) and a very strong positive correlation (r = 0.998) between the ground truth distance values and those obtained from the Radar Sensor.


Sensors ◽  
2018 ◽  
Vol 18 (11) ◽  
pp. 3705 ◽  
Author(s):  
Delaram Jarchi ◽  
Dario Salvi ◽  
Lionel Tarassenko ◽  
David Clifton

Respiratory rate (RR) is a key parameter used in healthcare for monitoring and predicting patient deterioration. However, continuous and automatic estimation of this parameter from wearable sensors is still a challenging task. Various methods have been proposed to estimate RR from wearable sensors using windowed segments of the data; e.g., often using a minimum of 32 s. Little research has been reported in the literature concerning the instantaneous detection of respiratory rate from such sources. In this paper, we develop and evaluate a method to estimate instantaneous respiratory rate (IRR) from body-worn reflectance photoplethysmography (PPG) sensors. The proposed method relies on a nonlinear time-frequency representation, termed the wavelet synchrosqueezed transform (WSST). We apply the latter to derived modulations of the PPG that arise from the act of breathing.We validate the proposed algorithm using (i) a custom device with a PPG probe placed on various body positions and (ii) a commercial wrist-worn device (WaveletHealth Inc., Mountain View, CA, USA). Comparator reference data were obtained via a thermocouple placed under the nostrils, providing ground-truth information concerning respiration cycles. Tracking instantaneous frequencies was performed in the joint time-frequency spectrum of the (4 Hz re-sampled) respiratory-induced modulation using the WSST, from data obtained from 10 healthy subjects. The estimated instantaneous respiratory rates have shown to be highly correlated with breath-by-breath variations derived from the reference signals. The proposed method produced more accurate results compared to averaged RR obtained using 32 s windows investigated with overlap between successive windows of (i) zero and (ii) 28 s. For a set of five healthy subjects, the averaged similarity between reference RR and instantaneous RR, given by the longest common subsequence (LCSS) algorithm, was calculated as 0.69; this compares with averaged similarity of 0.49 using 32 s windows with 28 s overlap between successive windows. The results provide insight into estimation of IRR and show that upper body positions produced PPG signals from which a better respiration signal was extracted than for other body locations.


2021 ◽  
Vol 4 (1) ◽  
Author(s):  
Ananyananda Dasari ◽  
Sakthi Kumar Arul Prakash ◽  
László A. Jeni ◽  
Conrad S. Tucker

AbstractThis work investigates the estimation biases of remote photoplethysmography (rPPG) methods for pulse rate measurement across diverse demographics. Advances in photoplethysmography (PPG) and rPPG methods have enabled the development of contact and noncontact approaches for continuous monitoring and collection of patient health data. The contagious nature of viruses such as COVID-19 warrants noncontact methods for physiological signal estimation. However, these approaches are subject to estimation biases due to variations in environmental conditions and subject demographics. The performance of contact-based wearable sensors has been evaluated, using off-the-shelf devices across demographics. However, the measurement uncertainty of rPPG methods that estimate pulse rate has not been sufficiently tested across diverse demographic populations or environments. Quantifying the efficacy of rPPG methods in real-world conditions is critical in determining their potential viability as health monitoring solutions. Currently, publicly available face datasets accompanied by physiological measurements are typically captured in controlled laboratory settings, lacking diversity in subject skin tones, age, and cultural artifacts (e.g, bindi worn by Indian women). In this study, we collect pulse rate and facial video data from human subjects in India and Sierra Leone, in order to quantify the uncertainty in noncontact pulse rate estimation methods. The video data are used to estimate pulse rate using state-of-the-art rPPG camera-based methods, and compared against ground truth measurements captured using an FDA-approved contact-based pulse rate measurement device. Our study reveals that rPPG methods exhibit similar biases when compared with a contact-based device across demographic groups and environmental conditions. The mean difference between pulse rates measured by rPPG methods and the ground truth is found to be ~2% (1 beats per minute (b.p.m.)), signifying agreement of rPPG methods with the ground truth. We also find that rPPG methods show pulse rate variability of ~15% (11 b.p.m.), as compared to the ground truth. We investigate factors impacting rPPG methods and discuss solutions aimed at mitigating variance.


2019 ◽  
Vol 19 (17) ◽  
pp. 7583-7590
Author(s):  
Yi Chiew Han ◽  
Kiing Ing Wong ◽  
Iain Murray

Sensors ◽  
2018 ◽  
Vol 18 (11) ◽  
pp. 3811 ◽  
Author(s):  
Marco Avvenuti ◽  
Nicola Carbonaro ◽  
Mario Cimino ◽  
Guglielmo Cola ◽  
Alessandro Tognetti ◽  
...  

Wearable sensors may enable the continuous monitoring of gait out of the clinic without requiring supervised tests and costly equipment. This paper investigates the use of a single wearable accelerometer to detect foot contact times and estimate temporal gait parameters (stride time, swing and stance duration). The experiments considered two possible body positions for the accelerometer: over the lower trunk and inside a trouser pocket. The latter approach could be implemented using a common smartphone. Notably, during the experiments, the ground truth was obtained by using a pair of sensorized shoes. Unlike ambient sensors and camera-based systems, sensorized shoes enable the evaluation of body-worn sensors even during longer walks. Experiments showed that both trunk and pocket positions achieved promising results in estimating gait parameters, with a mean absolute error below 50 ms.


Sensors ◽  
2020 ◽  
Vol 20 (3) ◽  
pp. 747 ◽  
Author(s):  
Ines Sorrentino ◽  
Francisco Javier Andrade Chavez ◽  
Claudia Latella ◽  
Luca Fiorio ◽  
Silvio Traversaro ◽  
...  

Wearable sensors are gaining in popularity because they enable outdoor experimental monitoring. This paper presents a cost-effective sensorised insole based on a mesh of tactile capacitive sensors. Each sensor’s spatial resolution is about 4 taxels/cm 2 in order to have an accurate reconstruction of the contact pressure distribution. As a consequence, the insole provides information such as contact forces, moments, and centre of pressure. To retrieve this information, a calibration technique that fuses measurements from a vacuum chamber and shoes equipped with force/torque sensors is proposed. The validation analysis shows that the best performance achieved a root mean square error (RMSE) of about 7   N for the contact forces and 2   N m for the contact moments when using the force/torque shoe data as ground truth. Thus, the insole may be an alternative to force/torque sensors for certain applications, with a considerably more cost-effective and less invasive hardware.


2021 ◽  
Vol 4 (1) ◽  
Author(s):  
Robert Amelard ◽  
Eric T. Hedge ◽  
Richard L. Hughson

AbstractOxygen consumption ($$\dot{\,{{\mbox{V}}}}{{{\mbox{O}}}}_{2}$$ V ̇ O 2 ) provides established clinical and physiological indicators of cardiorespiratory function and exercise capacity. However, $$\dot{\,{{\mbox{V}}}}{{{\mbox{O}}}}_{2}$$ V ̇ O 2 monitoring is largely limited to specialized laboratory settings, making its widespread monitoring elusive. Here we investigate temporal prediction of $$\dot{\,{{\mbox{V}}}}{{{\mbox{O}}}}_{2}$$ V ̇ O 2 from wearable sensors during cycle ergometer exercise using a temporal convolutional network (TCN). Cardiorespiratory signals were acquired from a smart shirt with integrated textile sensors alongside ground-truth $$\dot{\,{{\mbox{V}}}}{{{\mbox{O}}}}_{2}$$ V ̇ O 2 from a metabolic system on 22 young healthy adults. Participants performed one ramp-incremental and three pseudorandom binary sequence exercise protocols to assess a range of $$\dot{\,{{\mbox{V}}}}{{{\mbox{O}}}}_{2}$$ V ̇ O 2 dynamics. A TCN model was developed using causal convolutions across an effective history length to model the time-dependent nature of $$\dot{\,{{\mbox{V}}}}{{{\mbox{O}}}}_{2}$$ V ̇ O 2 . Optimal history length was determined through minimum validation loss across hyperparameter values. The best performing model encoded 218 s history length (TCN-VO2 A), with 187, 97, and 76 s yielding <3% deviation from the optimal validation loss. TCN-VO2 A showed strong prediction accuracy (mean, 95% CI) across all exercise intensities (−22 ml min−1, [−262, 218]), spanning transitions from low–moderate (−23 ml min−1, [−250, 204]), low–high (14 ml min−1, [−252, 280]), ventilatory threshold–high (−49 ml min−1, [−274, 176]), and maximal (−32 ml min−1, [−261, 197]) exercise. Second-by-second classification of physical activity across 16,090 s of predicted $$\dot{\,{{\mbox{V}}}}{{{\mbox{O}}}}_{2}$$ V ̇ O 2 was able to discern between vigorous, moderate, and light activity with high accuracy (94.1%). This system enables quantitative aerobic activity monitoring in non-laboratory settings, when combined with tidal volume and heart rate reserve calibration, across a range of exercise intensities using wearable sensors for monitoring exercise prescription adherence and personal fitness.


Sign in / Sign up

Export Citation Format

Share Document