A Posture Recognition-Based Fall Detection System for Monitoring an Elderly Person in a Smart Home Environment
read more
Citations
Internet of Things: Architectures, Protocols, and Applications
Smart Homes for Elderly Healthcare—Recent Advances and Research Challenges
A Survey on Activity Detection and Classification Using Wearable Sensors
Remote patient monitoring: a comprehensive study
Survey on Fall Detection and Fall Prevention Using Wearable and External Sensors
References
Statistical learning theory
Pattern Recognition and Machine Learning
Pattern Recognition and Machine Learning
Adaptive background mixture models for real-time tracking
Pfinder: real-time tracking of the human body
Related Papers (5)
Frequently Asked Questions (12)
Q2. How many people were recorded to form their posture dataset?
To form their posture dataset, 3200 postures (including 800 stands, 800 sits, 800 lies and 800 bends) from 15 people were recorded.
Q3. What are the main problems of computer vision based fall detection systems?
in most computer vision based fall detection systems, only the alarm signal (sometimes with a short video clip as further confirmation of whether an elderly person has fallen or not) will be sent to the caregivers or family members when a fall is detected; additionally, the original video recordings of an elderly person’s normal activities will not be stored, nor transmitted.
Q4. What is the common approach for detecting moving objects from the background?
In visual surveillance, a common approach for discriminating moving objects from the background is detection by background subtraction.
Q5. How many images are used in the training dataset?
Assuming the training dataset The authorcontains a number of N images: The author= {imag1, ..., imagN}, then, for a single pixel (x,y), it has N training samples imag(x, y)1, ..., imag(x, y)N .
Q6. What are the advantages of the proposed fall detection system?
(2) If the training dataset is large enough, the well-trained classifier can effectively distinguish different types of postures, which are used for fall detection.
Q7. What are the features of the fitted ellipse and projection histogram?
From the extracted foreground silhouette, the authors extract features from the fitted ellipse and projection histogram, which are used for classification purposes.
Q8. What are the problems that can be addressed by using a multiple cameras scheme?
as the authors have discussed, multiple moving objects and occlusions are two problems needed to be solved for their fall detection system, which can be addressed by using a multiple cameras scheme with adding corresponding modules for people counting and object classification.
Q9. What is the classification method for a lying on a sofa?
For (b), although a ‘lie’ posture is detected, the human body blob is not in the floor region, so the lying on the sofa case is correctly classified as non-fall.
Q10. What are the main problems of non-computer vision based methods?
Although non-computer vision based methods may appear to be suitable for wide application in the fall detection field, several problems do exist; they are either inconvenient (elderly people have to wear acceleration sensors) or easily affected by noise in the environment (acoustic sensors and floor vibration sensors).
Q11. What are the two types of methods proposed for detecting falls?
Different methods have been proposed for detecting falls and are mainly divided into two categories: non-computer vision based methods and computer vision based methods.
Q12. What are the two types of postures that are not detected as falls?
For (d) and (e), either the detected ‘bend’ posture does not hold for a long time (for case (d), a person ties his shoe lace and the ‘bend’ posture recovers to ‘stand’ posture in a short time), or the posture is not in the ground region (only a small portion of the human body region is in the ground), so they are not detected as falls.