Dataset Information

Robust Self-Adaptation Fall-Detection System Based on Camera Height.

ABSTRACT: Vision-based fall-detection methods have been previously studied but many have limitations in terms of practicality. Due to differences in rooms, users do not set the camera or sensors at the same height. However, few studies have taken this into consideration. Moreover, some fall-detection methods are lacking in terms of practicality because only standing, sitting and falling are taken into account. Hence, this study constructs a data set consisting of various daily activities and fall events and studies the effect of camera/sensor height on fall-detection accuracy. Each activity in the data set is carried out by eight participants in eight directions and taken with the depth camera at five different heights. Many related studies heavily depended on human segmentation by using Kinect SDK but this is not reliable enough. To address this issue, this study proposes Enhanced Tracking and Denoising Alex-Net (ETDA-Net) to improve tracking and denoising performance and classify fall and non-fall events. Experimental results indicate that fall-detection accuracy is affected by camera height, against which ETDA-Net is robust, outperforming traditional deep learning based fall-detection methods.

SUBMITTER: Kong X

PROVIDER: S-EPMC6749320 | biostudies-literature | 2019 Aug

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Robust Self-Adaptation Fall-Detection System Based on Camera Height.

Kong Xiangbo X Chen Lehan L Wang Zhichen Z Chen Yuxi Y Meng Lin L Tomiyama Hiroyuki H

Sensors (Basel, Switzerland) 20190830 17

Vision-based fall-detection methods have been previously studied but many have limitations in terms of practicality. Due to differences in rooms, users do not set the camera or sensors at the same height. However, few studies have taken this into consideration. Moreover, some fall-detection methods are lacking in terms of practicality because only standing, sitting and falling are taken into account. Hence, this study constructs a data set consisting of various daily activities and fall events a ...[more]

PMID: 31480384

Similar Datasets

Project description:BACKGROUND:Accidental falls are a common cause of injury and deaths. Both ground-level falls (GLF) and non-GLF may lead to significant morbidity or mortality. This study aimed to explore the relationship between height of falls and mortality. METHOD:This is a retrospective study based on the data from a registered trauma database and included 8699 adult patients who were hospitalized between 1 January 2009 and 31 December 2017 for the treatment of fall-related injuries. Study subjects were divided into three groups of two categories based on the height of fall: GLF (group I: < 1 m) and non-GLF (group II: 1-6 m and group III: > 6 m). The primary outcome was in-hospital mortality. The adjusted odds ratio (AOR) of mortality adjusted for age, sex, and comorbidities with or without an injury severity score (ISS) was calculated using multiple logistic regression. RESULTS:Among the 7001 patients in group I, 1588 in group II, and 110 in group III, patients in the GLF group were older, predominantly female, had less intentional injuries, and had more pre-existing comorbidities than those in the non-GLF group. The patients in the non-GLF group had a significantly lower Glasgow Coma Scale (GCS), a higher injury severity score (ISS), worse physiological responses, and required more procedures performed in the emergency department. The mortality rate for the patients in group I, II, and III were 2.5%, 3.5%, and 5.5%, respectively. After adjustment by age, sex, and comorbidities, group II and group III patients had significantly higher adjusted odds of mortality than group I patients (AOR 2.2, 95% CI 1.64-2.89, p < 0.001 and AOR 2.5, 95% CI 1.84-3.38, p < 0.001, respectively). With additional adjustment by ISS, group II did not have significantly higher adjusted odds of mortality than group I patients (AOR 1.4, 95% CI 0.95-2.22, p = 0.082), but group III patients still had significantly higher adjusted odds of mortality than group I patients (AOR 10.0, 95% CI 2.22-33.33, p = 0.002). CONCLUSION:This study suggested that patients who sustained GLF and non-GLF were distinct groups of patients, and the height of fall did have an impact on mortality in patients of fall accidents. A significantly higher adjusted odds of mortality was found in the GLF group than in the non-GLF group after adjusting for age, sex, and comorbidities.

Dataset Information

Robust Self-Adaptation Fall-Detection System Based on Camera Height.

Publications

Robust Self-Adaptation Fall-Detection System Based on Camera Height.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets