a presentation and a report on the topic of recognizing human activities from sensors using soft computing techniques
Posts: 14,118
Threads: 61
Joined: Oct 2014
Services in the ubiquitous environment have increased. These types of services focus on the context of user activities, location or environment. There were many studies on the recognition of these contexts using various sensory resources. To recognize human activity, many of them used an accelerometer, which shows good accuracy to recognize the activities of the user's movements, but did not recognize stable activities that could be classified by the user's emotion and inferred by physiological sensors. In this article, we take advantage of multiple sensor signals to recognize user activity. As Bramband includes an accelerometer and physiological sensors, we use them with a fuzzy Bayesian network for continuous sensor data. The diffuse membership function uses three stages differentiated by the distribution of each of the sensor data. Experiments on the accuracy of activity recognition have been carried out by combining the uses of accelerometers and physiological signals. For the result, the total accuracy appears to be 74.4% for activities that include dynamic activities and stable activities, using physiological signals and a 2-axis accelerometer. When we use only the physiological signals the accuracy is 60.9%, and when we use the 2-axis accelerometer the precision is 44.2%. We show that the use of physiological signals with accelerometer is more efficient in the recognition of activities.