China
Africa
Explore chapters and articles related to this topic
Advanced Sensors for Mechatronics
Published in Marina Indri, Roberto Oboe, Mechatronics and Robotics, 2020
Naoki Oda, Toshiaki Tsuji, Yasue Mitsukura, Takahiro Ishikawa, Yuta Tawaki, Toshiyuki Murakami, Satoshi Suzuki, Yasutaka Fujimoto
Activity Recognition is a generic term for technologies to identify human activities and actions from time-series data measured by various types of sensors such as acceleration sensor, camera, ultrasonic sensor, laser-range finder, depth sensor, and Kinect. Most basic AR is for acceleration sensors attached to the human body, and such AR began in the 2000s [48]. Since then, various types have been studied, such as a simple detection of ambulation [49], discrimination of several activities [50], and a precise action discrimination using a smart tag system [51]. Here, AR for children is introduced. This AR is a part of the functions of Kinder-GuaRdian system which is a child–parents–childminder support system in kindergarten, and it consists of a sensor node attached to the child, interactive measurement robots, and child-care data analyzers [52]. Recording the child’s activity and finding his/her life rhythm, their information is utilized for his/her health care and growth management. Recently, in advanced countries, the rete of child obesity has been increasing [53], and it is said that childhood obesity causes lifestyle diseases when they become adults [54]; hence, such child AR may become important to national healthcare.
Scaling Smart Environments
Published in Mohammad Ilyas, Sami S. Alwakeel, Mohammed M. Alwakeel, el-Hadi M. Aggoune, Sensor Networks for Sustainable Development, 2017
Intelligent systems that focus on the needs of a human require information about the activities being performed by the human. At the core of these systems, then, is activity recognition, which is a challenging and well-researched problem. Sensors in a smart home generate events that consist of a date, a time, a sensor identifier, and a sensor message. The generally accepted approach to activity recognition is to design and/or use machine learning techniques to map a sequence of sensor data to a corresponding activity label. Online activity recognition, or recognizing activities in real time from streaming data, introduces challenges that do not occur in the case of offline learning with presegmented data. However, this is an approach to activity recognition that needs to be considered in order to scale the capabilities of smart environments.
Sensor Networking Software and Architectures
Published in John R. Vacca, Handbook of Sensor Networking, 2015
Smartphones have already enabled a plethora of mobile sensing applications (Abdelzaher et al., 2007; Campbell et al., 2006; Honicky et al., 2008; Mohan et al., 2008) in gaming, smart environments, surveillance, emergency response, and social networks. Specially, activity recognition through mobile sensing and wearable sensors has led to many health-care applications, such as fitness monitoring, elder care support, and cognitive assistance (Choudhury et al., 2008). The expanding sensing capabilities of mobile phones have gone beyond the sensor networks' focus on environmental and infrastructure monitoring where people are now the carriers of sensing devices, the sources, and the consumers of sensed events (Azizyan et al., 2009; Kansal et al., 2007; Lu et al., 2009; Miluzzo et al., 2008; Siewiorek et al., 2003).
Deep Maxout Network for human action and abnormality detection using Chronological Poor and Rich Optimization
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2023
Human activity recognition is a process of differentiating different human activities that depend upon the shape, size and various posture exhibited by the humans while doing a specific action. Human activity recognition in a complicated environment causes significant impacts in the performance as well as in the accuracy and it still exists as a challenging task in human action detection. To cope up with this need, an effectual strategy for human action abnormality recognition named CPRO-based Deep Maxout Network is designed. In this research, the process involves mainly three phases, such as frame, feature extraction and human action recognition. Moreover, human movement and abnormality recognition are detected using Deep Maxout Network, which is trained utilising developed CPRO. The developed CPRO is devised by applying Chronological concept to the PRO algorithm. Moreover, the proposed approach shows the high accuracy of 0.954, high sensitivity of 0.958 and high specificity of 0.960 for abnormality recognition. The future enhancement would be the inclusion of various optimisation algorithms in order to boost up the training process of classifiers that would enhance the overall performance of the system.
Efficient key frame extraction and hybrid wavelet convolutional manta ray foraging for sports video classification
Published in The Imaging Science Journal, 2023
Moreover, the proposed method achieves 90% recall and precision value with less detection time. Human activity recognition is considered a very challenging issue which requires recognition of the activity performed by a group of people or single individuals from spatiotemporal data. A solution must be essential for this kind of problem in computer vision. ConvLSTM framework is considered one of the solutions for this problem, and many spatiotemporal computer vision applications utilize this architecture frequently. In this research, Sarah Khater et al. [36] 2022 developed a new layer named residual inception convolutional recurrent layer, ResIncConvLSTM, a variant of the ConvLSTM layer. Compared to the ConvLSTM baseline architecture, the proposed method obtains 7% more improvement in accuracy performance. Table 1 lists the existing literature and its merits and demerits.
Use of smartphone sensors to quantify the productive cycle elements of hand fallers on industrial cable logging operations
Published in International Journal of Forest Engineering, 2019
Robert F. Keefe, Eloise G. Zimbelman, Ann M. Wempe
Sensor-based activity recognition uses data collected about individuals and their environment to characterize physical activities. Activity recognition is a central technique in big data science and underlies the concept of the quantified self (Swan 2013). Activity recognition using sensors based on the body or on objects is used in personal fitness applications (Ermes et al. 2008), health care (Anjum and Ilyas 2013; Lau et al. 2010), and the development of smart homes (Gu et al. 2009; Chen et al. 2012). Both smartphones and smartwatches are used widely in activity recognition, as many of these devices are now equipped with a variety of sensors, including accelerometers, gyroscopes, barometers, thermometers, decibel meters (microphones), magnetometers, lux meters, optical heart rate sensors, and GNSS chips (Anjum and Ilyas 2013; Trost et al. 2014; del Rosario et al. 2015; Shoaib et al. 2015, 2016; Weiss et al. 2016). These sensors can identify position, movement (Anjum and Ilyas 2013; Bayat et al. 2014), physiological indicators (Chen et al. 2012), and environmental characteristics (Gu et al. 2009), which in turn can be used as the basis for predictive models.