Human Activity Recognition for Physical Rehabilitation

Leightley, Daniel; Darby, John; Li, Baihua; McPhee, Jamie S.; Yap, Moi Hoon

doi:10.1109/smc.2013.51

Cited by 36 publications

(16 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…A random sample of participant's from the K3Da dataset are extracted, and each joint group is modelled using an SVM with 10-fold cross-validation, Figure 6 demonstrates the training and evaluation pipeline. While it is possible to train a single SVM, indeed [1], [42] obtained high accuracy results for the task of recognition it is not suitable for clinical scenarios. If we follow these type of approaches, subtle motion variations would be overshadowed resulting in over generalisation (over fitting) leading to inter-/intra-class confusion between "good" mobility and "poor" mobility.…”

Section: Machine Learning Parameters Requiredmentioning

confidence: 99%

“…1) Labelling and Computation of Human Mobility: For a typical recognition task, prior knowledge of the class label is required. This is usually straightforward to determine, for example a person walking or jumping can easily be defined with a single label [1]. However, the task becomes very difficult to identify and label in the context of different styles of the same motion.…”

Section: Motion Analysismentioning

confidence: 99%

See 1 more Smart Citation

Automated Analysis and Quantification of Human Mobility Using a Depth Sensor

Leightley

McPhee

Yap

2017

IEEE J. Biomed. Health Inform.

Self Cite

View full text Add to dashboard Cite

Analysis and quantification of human motion to support clinicians in the decision-making process is the desired outcome for many clinical-based approaches. However, generating statistical models that are free from human interpretation and yet representative is a difficult task. In this paper, we propose a framework that automatically recognizes and evaluates human mobility impairments using the Microsoft Kinect One depth sensor. The framework is composed of two parts. First, it recognizes motions, such as sit-to-stand or walking 4 m, using abstract feature representation techniques and machine learning. Second, evaluation of the motion sequence in the temporal domain by comparing the test participant with a statistical mobility model, generated from tracking movements of healthy people. To complement the framework, we propose an automatic method to enable a fairer, unbiased approach to label motion capture data. Finally, we demonstrate the ability of the framework to recognize and provide clinically relevant feedback to highlight mobility concerns, hence providing a route toward stratified rehabilitation pathways and clinician-led interventions.

show abstract

Section: Machine Learning Parameters Requiredmentioning

confidence: 99%

Section: Motion Analysismentioning

confidence: 99%

Automated Analysis and Quantification of Human Mobility Using a Depth Sensor

Leightley

McPhee

Yap

2017

IEEE J. Biomed. Health Inform.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Support Vector Machines [3] is another supervised learning algorithm which finds the optimal separating hyperplane to decide where to classify data. Both RF and SVM have been used in facial expression recognition [20,14,25,23] and also in other methods such as physical rehabilitation [9] and bioinformatics [19].…”

Section: Related Workmentioning

confidence: 99%

Micro-Facial Movements: An Investigation on Spatio-Temporal Descriptors

Davison

Yap

Costen

et al. 2015

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

This paper aims to investigate whether micro-facial movement sequences can be distinguished from neutral face sequences. As a micro-facial movement tends to be very quick and subtle, classifying when a movement occurs compared to the face without movement can be a challenging computer vision problem. Using local binary patterns on three orthogonal planes and Gaussian derivatives, local features, when interpreted by machine learning algorithms, can accurately describe when a movement and non-movement occurs. This method can then be applied to help aid humans in detecting when the small movements occur. This also differs from current literature as most only concentrate in emotional expression recognition. Using the CASME II dataset, the results from the investigation of different descriptors have shown a higher accuracy compared to state-of-the-art methods.

show abstract

“…It incorporates multiple high-resolution cameras and makes use of infra-red reflective markers to achieve millimetre resolution of 3-D spacial displacements at greater than 100 frames per second 7,11 . Physiotherapists and a Medical student who had received specific training in shoulder ROM assessment for this project.…”

Section: Resultsmentioning

confidence: 99%