The Experts below are selected from a list of 12033 Experts worldwide ranked by ideXlab platform
Behzad Dariush - One of the best experts on this subject based on the ideXlab platform.
-
spatio temporal pyramid graph convolutions for human action recognition and Postural Assessment
Workshop on Applications of Computer Vision, 2020Co-Authors: Behnoosh Parsa, Athma Narayanan, Behzad DariushAbstract:Recognition of human actions and associated interactions with objects and the environment is an important problem in computer vision due to its potential applications in a variety of domains. Recently, graph convolutional networks that extract features from the skeleton have demonstrated promising performance. In this paper, we propose a novel Spatio-Temporal Pyramid Graph Convolutional Network (ST-PGN) for online action recognition for ergonomics risk Assessment that enables the use of features from all levels of the skeleton feature hierarchy. The proposed algorithm outperforms state-of-art action recognition algorithms tested on two public benchmark datasets typically used for Postural Assessment (TUM and UW-IOM). We also introduce a pipeline to enhance Postural Assessment methods with online action recognition techniques. Finally, the proposed algorithm is integrated with a traditional ergonomics risk index (REBA) to demonstrate the potential value for Assessment of musculoskeletal disorders in occupational safety.
-
WACV - Spatio-Temporal Pyramid Graph Convolutions for Human Action Recognition and Postural Assessment
2020 IEEE Winter Conference on Applications of Computer Vision (WACV), 2020Co-Authors: Behnoosh Parsa, Athma Narayanan, Behzad DariushAbstract:Recognition of human actions and associated interactions with objects and the environment is an important problem in computer vision due to its potential applications in a variety of domains. Recently, graph convolutional networks that extract features from the skeleton have demonstrated promising performance. In this paper, we propose a novel Spatio-Temporal Pyramid Graph Convolutional Network (ST-PGN) for online action recognition for ergonomics risk Assessment that enables the use of features from all levels of the skeleton feature hierarchy. The proposed algorithm outperforms state-of-art action recognition algorithms tested on two public benchmark datasets typically used for Postural Assessment (TUM and UW-IOM). We also introduce a pipeline to enhance Postural Assessment methods with online action recognition techniques. Finally, the proposed algorithm is integrated with a traditional ergonomics risk index (REBA) to demonstrate the potential value for Assessment of musculoskeletal disorders in occupational safety.
-
Spatio-Temporal Pyramid Graph Convolutions for Human Action Recognition and Postural Assessment
arXiv: Computer Vision and Pattern Recognition, 2019Co-Authors: Behnoosh Parsa, Athma Narayanan, Behzad DariushAbstract:Recognition of human actions and associated interactions with objects and the environment is an important problem in computer vision due to its potential applications in a variety of domains. The most versatile methods can generalize to various environments and deal with cluttered backgrounds, occlusions, and viewpoint variations. Among them, methods based on graph convolutional networks that extract features from the skeleton have demonstrated promising performance. In this paper, we propose a novel Spatio-Temporal Pyramid Graph Convolutional Network (ST-PGN) for online action recognition for ergonomic risk Assessment that enables the use of features from all levels of the skeleton feature hierarchy. The proposed algorithm outperforms state-of-art action recognition algorithms tested on two public benchmark datasets typically used for Postural Assessment (TUM and UW-IOM). We also introduce a pipeline to enhance Postural Assessment methods with online action recognition techniques. Finally, the proposed algorithm is integrated with a traditional ergonomic risk index (REBA) to demonstrate the potential value for Assessment of musculoskeletal disorders in occupational safety.
Behnoosh Parsa - One of the best experts on this subject based on the ideXlab platform.
-
spatio temporal pyramid graph convolutions for human action recognition and Postural Assessment
Workshop on Applications of Computer Vision, 2020Co-Authors: Behnoosh Parsa, Athma Narayanan, Behzad DariushAbstract:Recognition of human actions and associated interactions with objects and the environment is an important problem in computer vision due to its potential applications in a variety of domains. Recently, graph convolutional networks that extract features from the skeleton have demonstrated promising performance. In this paper, we propose a novel Spatio-Temporal Pyramid Graph Convolutional Network (ST-PGN) for online action recognition for ergonomics risk Assessment that enables the use of features from all levels of the skeleton feature hierarchy. The proposed algorithm outperforms state-of-art action recognition algorithms tested on two public benchmark datasets typically used for Postural Assessment (TUM and UW-IOM). We also introduce a pipeline to enhance Postural Assessment methods with online action recognition techniques. Finally, the proposed algorithm is integrated with a traditional ergonomics risk index (REBA) to demonstrate the potential value for Assessment of musculoskeletal disorders in occupational safety.
-
WACV - Spatio-Temporal Pyramid Graph Convolutions for Human Action Recognition and Postural Assessment
2020 IEEE Winter Conference on Applications of Computer Vision (WACV), 2020Co-Authors: Behnoosh Parsa, Athma Narayanan, Behzad DariushAbstract:Recognition of human actions and associated interactions with objects and the environment is an important problem in computer vision due to its potential applications in a variety of domains. Recently, graph convolutional networks that extract features from the skeleton have demonstrated promising performance. In this paper, we propose a novel Spatio-Temporal Pyramid Graph Convolutional Network (ST-PGN) for online action recognition for ergonomics risk Assessment that enables the use of features from all levels of the skeleton feature hierarchy. The proposed algorithm outperforms state-of-art action recognition algorithms tested on two public benchmark datasets typically used for Postural Assessment (TUM and UW-IOM). We also introduce a pipeline to enhance Postural Assessment methods with online action recognition techniques. Finally, the proposed algorithm is integrated with a traditional ergonomics risk index (REBA) to demonstrate the potential value for Assessment of musculoskeletal disorders in occupational safety.
-
Spatio-Temporal Pyramid Graph Convolutions for Human Action Recognition and Postural Assessment
arXiv: Computer Vision and Pattern Recognition, 2019Co-Authors: Behnoosh Parsa, Athma Narayanan, Behzad DariushAbstract:Recognition of human actions and associated interactions with objects and the environment is an important problem in computer vision due to its potential applications in a variety of domains. The most versatile methods can generalize to various environments and deal with cluttered backgrounds, occlusions, and viewpoint variations. Among them, methods based on graph convolutional networks that extract features from the skeleton have demonstrated promising performance. In this paper, we propose a novel Spatio-Temporal Pyramid Graph Convolutional Network (ST-PGN) for online action recognition for ergonomic risk Assessment that enables the use of features from all levels of the skeleton feature hierarchy. The proposed algorithm outperforms state-of-art action recognition algorithms tested on two public benchmark datasets typically used for Postural Assessment (TUM and UW-IOM). We also introduce a pipeline to enhance Postural Assessment methods with online action recognition techniques. Finally, the proposed algorithm is integrated with a traditional ergonomic risk index (REBA) to demonstrate the potential value for Assessment of musculoskeletal disorders in occupational safety.
Carina U. Persson - One of the best experts on this subject based on the ideXlab platform.
-
Inter-rater reliability of the Swedish modified version of the Postural Assessment Scale for Stroke Patients (SwePASS) in the acute phase after stroke.
Topics in Stroke Rehabilitation, 2019Co-Authors: Gunilla M. Bergqvist, Salmir Nasic, Carina U. PerssonAbstract:Background: Before implementation of the new scale, the Swedish modified version of the Postural Assessment Scale for Stroke Patients (SwePASS), to clinical practice, it is fundamental to analyze i...
-
Measurement properties of the Swedish modified version of the Postural Assessment Scale for Stroke Patients (SwePASS) using Rasch analysis.
European Journal of Physical and Rehabilitation Medicine, 2017Co-Authors: Carina U. Persson, Annika Linder, Peter HagellAbstract:BACKGROUND: A previous small-sample (N.=150) Rasch analysis of the Swedish modified version of the Postural Assessment Scale for Stroke Patients (SwePASS) suggested problems regarding response categories and redundant items that need confirmation in larger samples with more severe strokes. AIM: The aim of this study was to evaluate the measurement properties of the SwePASS in patients with acute stroke. DESIGN: A multicenter, cross-sectional study. SETTING: Two stroke units in Western Sweden. POPULATION: The study cohort included 250 consecutive inpatients undergoing rehabilitation after acute stroke. METHODS: The SwePASS Assessments were performed once within the first four days after admission to the stroke units. The data were analyzed according to the Rasch measurement model regarding targeting, model fit, reliability, response category function, local dependence and differential item functioning. RESULTS: Postural control of 250 patients (median age, 76.5 years) was assessed with the SwePASS within median of two days after admission to the stroke units. The SwePASS covered a continuum of different levels of Postural control, but had suboptimal targeting with insufficient representation of lower and higher levels of Postural control. The reliability was high, the item fit statistics were generally acceptable and there was no differential item functioning by sex, age and stroke localization. However, response categories did not function as expected for four of the 12 SwePASS items and five items exhibited local dependency. CONCLUSIONS: The SwePASS exhibited several promising measurement properties. To improve the scale, poor targeting, illogical response categories and local dependency should be addressed. CLINICAL REHABILITATION IMPACT: The SwePASS provides valuable clinical information regarding Postural control in the acute phase after stroke.
-
Rasch analysis of the modified version of the Postural Assessment Scale for Stroke patients: Postural Stroke Study in Gothenburg (POSTGOT)
BMC Neurology, 2014Co-Authors: Carina U. Persson, Katharina S. Sunnerhagen, Åsa Lundgren-nilssonAbstract:Background The modified version of the Postural Assessment Scale for Stroke Patients (SwePASS) is a new ordinal outcome measurement designed to assess Postural control in patients with stroke. Before implementation of SwePASS into the clinical setting, it is necessary to know its measurement properties. Thus, the aim of the study was to evaluate the measurement properties of the SwePASS.
-
Responsiveness of a modified version of the Postural Assessment scale for stroke patients and longitudinal change in Postural control after stroke- Postural Stroke Study in Gothenburg (POSTGOT) -
Journal of Neuroengineering and Rehabilitation, 2013Co-Authors: Carina U. Persson, Anna Danielsson, Katharina S. Sunnerhagen, Anna Grimby-ekman, Per-olof HanssonAbstract:Background: Responsiveness data certify that a change in a measurement output represents a real change, not a measurement error or biological variability. The objective was to evaluate the responsiveness of the modified version of the Postural Assessment Scale for Stroke Patients (SwePASS) in patients with a first event of stroke. An additional aim was to estimate the change in Postural control during the first 12 months after stroke onset. Methods: The SwePASS Assessments were conducted during the first week and 3, 6 and 12 months after stroke in 90 patients. Svensson’s method, Relative Position (RP), Relative Concentration (RC) and Relative Rank Variance (RV), were used to estimate the scale’s responsiveness and the patients’ change in Postural control over time. Results: From the first week to 3 months after stroke, the patients improved in terms of Postural control with 2 to 12 times larger systematic changes in Relative Position (RP), for which 9 items and the total score showed a significant responsiveness to change when compared to the intrarater reliability measurement error of the SwePASS reported in a previous study. When SwePASS was used to assess change in Postural control between the first week and 3 months, 74% of the patients received higher scores while 10% received lower scores, RP 0.31 (95% CI 0.219-0.402). The corresponding figures between 3 and 6 and between 6 and 12 months were 37% and 16%, RP 0.09 (95% CI 0.030-0.152), and 18% and 26%, RP �0.07 (95% CI �0.134- (�0.010)), respectively. Conclusions: The SwePASS is responsive to change. Postural control evaluated using the SwePASS showed an improvement during the first 6 months after stroke. The measurement property, in the form of responsiveness, shows that the SwePASS scoring method can be considered for use in rehabilitation when assessing Postural control in patients after stroke, especially during the first 3 months.
-
A validation study using a modified version of Postural Assessment Scale for Stroke Patients: Postural Stroke Study in Gothenburg (POSTGOT).
Journal of Neuroengineering and Rehabilitation, 2011Co-Authors: Carina U. Persson, Per-olof Hansson, Anna Danielsson, Katharina S. SunnerhagenAbstract:Background: A modified version of Postural Assessment Scale for Stroke Patients (PASS) was created with some changes in the description of the items and clarifications in the manual (e.g. much help was defined as support from 2 persons). The aim of this validation study was to assess intrarater and interrater reliability using this modified version of PASS, at a stroke unit, for patients in the acute phase after their first event of stroke. Methods: In the intrarater reliability study 114 patients and in the interrater reliability study 15 patients were examined twice with the test within one to 24 hours in the first week after stroke. Spearman’s rank correlation, Kappa coefficients, Percentage Agreement and the newer rank-invariant methods; Relative Position, Relative Concentration and Relative rank Variance were used for the statistical analysis. Results: For the intrarater reliability Spearman’s rank correlations were 0.88-0.98 and k were 0.70-0.93 for the individual items. Small, statistically significant, differences were found for two items regarding Relative Position and for one item regarding Relative Concentration. There was no Relative rank Variance for any single item. For the interrater reliability, Spearman’s rank correlations were 0.77-0.99 for individual items. For some items there was a possible, even if not proved, reliability problem regarding Relative Position and Relative Concentration. There was no Relative rank Variance for the single items, except for a small Relative rank Variance for one item. Conclusions: The high intrarater and interrater reliability shown for the modified Postural Assessment Scale for Stroke Patients, the Swedish version of Postural Assessment Scale for Stroke Patients, with traditional and newer statistical analyses, particularly for Assessments performed by the same rater, support the use of the Swedish version of Postural Assessment Scale for Stroke Patients, in the acute stage after stroke both in clinical and research settings. In addition, the Swedish version of Postural Assessment Scale for Stroke Patients was easy to apply and fast to administer in clinic.
Athma Narayanan - One of the best experts on this subject based on the ideXlab platform.
-
spatio temporal pyramid graph convolutions for human action recognition and Postural Assessment
Workshop on Applications of Computer Vision, 2020Co-Authors: Behnoosh Parsa, Athma Narayanan, Behzad DariushAbstract:Recognition of human actions and associated interactions with objects and the environment is an important problem in computer vision due to its potential applications in a variety of domains. Recently, graph convolutional networks that extract features from the skeleton have demonstrated promising performance. In this paper, we propose a novel Spatio-Temporal Pyramid Graph Convolutional Network (ST-PGN) for online action recognition for ergonomics risk Assessment that enables the use of features from all levels of the skeleton feature hierarchy. The proposed algorithm outperforms state-of-art action recognition algorithms tested on two public benchmark datasets typically used for Postural Assessment (TUM and UW-IOM). We also introduce a pipeline to enhance Postural Assessment methods with online action recognition techniques. Finally, the proposed algorithm is integrated with a traditional ergonomics risk index (REBA) to demonstrate the potential value for Assessment of musculoskeletal disorders in occupational safety.
-
WACV - Spatio-Temporal Pyramid Graph Convolutions for Human Action Recognition and Postural Assessment
2020 IEEE Winter Conference on Applications of Computer Vision (WACV), 2020Co-Authors: Behnoosh Parsa, Athma Narayanan, Behzad DariushAbstract:Recognition of human actions and associated interactions with objects and the environment is an important problem in computer vision due to its potential applications in a variety of domains. Recently, graph convolutional networks that extract features from the skeleton have demonstrated promising performance. In this paper, we propose a novel Spatio-Temporal Pyramid Graph Convolutional Network (ST-PGN) for online action recognition for ergonomics risk Assessment that enables the use of features from all levels of the skeleton feature hierarchy. The proposed algorithm outperforms state-of-art action recognition algorithms tested on two public benchmark datasets typically used for Postural Assessment (TUM and UW-IOM). We also introduce a pipeline to enhance Postural Assessment methods with online action recognition techniques. Finally, the proposed algorithm is integrated with a traditional ergonomics risk index (REBA) to demonstrate the potential value for Assessment of musculoskeletal disorders in occupational safety.
-
Spatio-Temporal Pyramid Graph Convolutions for Human Action Recognition and Postural Assessment
arXiv: Computer Vision and Pattern Recognition, 2019Co-Authors: Behnoosh Parsa, Athma Narayanan, Behzad DariushAbstract:Recognition of human actions and associated interactions with objects and the environment is an important problem in computer vision due to its potential applications in a variety of domains. The most versatile methods can generalize to various environments and deal with cluttered backgrounds, occlusions, and viewpoint variations. Among them, methods based on graph convolutional networks that extract features from the skeleton have demonstrated promising performance. In this paper, we propose a novel Spatio-Temporal Pyramid Graph Convolutional Network (ST-PGN) for online action recognition for ergonomic risk Assessment that enables the use of features from all levels of the skeleton feature hierarchy. The proposed algorithm outperforms state-of-art action recognition algorithms tested on two public benchmark datasets typically used for Postural Assessment (TUM and UW-IOM). We also introduce a pipeline to enhance Postural Assessment methods with online action recognition techniques. Finally, the proposed algorithm is integrated with a traditional ergonomic risk index (REBA) to demonstrate the potential value for Assessment of musculoskeletal disorders in occupational safety.
Gerard Urrutia Cuchi - One of the best experts on this subject based on the ideXlab platform.
-
the effect of additional core stability exercises on improving dynamic sitting balance and trunk control for subacute stroke patients a randomized controlled trial
Clinical Rehabilitation, 2016Co-Authors: Rosa Cabanasvaldes, Caritat Bagurcalafat, Montserrat Girabentfarres, Fernanda Ma Caballerogomez, Montserrat Hernandezvalino, Gerard Urrutia CuchiAbstract:Objective:To examine the effect of core stability exercises on trunk control, dynamic sitting and standing balance, gait, and activities of daily living in subacute stroke patients.Design:A randomized controlled trial.Setting:Inpatient rehabilitation hospital in two centres.Subjects:Eighty patients (mean of 23.25 (±16.7) days post-stroke) were randomly assigned to an experimental group and a control group.Interventions:Both groups underwent conventional therapy for five days/week for five weeks and the experimental group performed core stability exercises for 15 min/day. The patients were assessed before and after intervention.Main measures:The Trunk Impairment Scale (Spanish-Version) and Function in Sitting Test were used to measure the primary outcome of dynamic sitting balance. Secondary outcome measures were standing balance and gait as evaluated via Berg Balance Scale, Tinetti Test, Brunel Balance Assessment, Postural Assessment Scale for Stroke (Spanish-Version), and activities of daily living using...