Sound waves bypass visual limitations to recognize human activity

[ad_1]

Video cameras continue to be widely used to monitor human activities in the areas of surveillance, health care, the home, and so on. Acoustic waves, such as sounds and other forms of vibration, are an alternative way of circumventing these limitations.

Unlike electromagnetic waves, such as those used in radar, acoustic waves can be used not only to search for objects, but also to identify them. As described in a new document published in the May 28 issue of Applied Physics Letters, the researchers used a two-dimensional acoustic network and convolutional neural networks to detect and analyze the sounds of human activity and identify these activities.

"If the accuracy of the identification is high enough, a lot of applications could be implemented," said Xinhua Guo, associate professor at Wuhan University of Technology. "For example, a medical alarm system could be activated if a person falls home and is detected, so immediate help could be provided and little personal information would be protected at the same time. "

Using a two-dimensional acoustic network of 256 receivers and four ultrasound emitters, the researchers were able to collect data on four different human activities: sitting, standing, walking and falling. They used a 40 kilohertz signal to bypass any potential contamination from the noise of a room and distinguish it from the sounds of identification.

Their tests reached an overall accuracy of 97.5% for time domain data and 100% for frequency domain data. The scientists also tested networks with fewer receivers (eight and four) and found that they produced results with less accuracy of human activity.

Guo said that acoustic systems are a better detection device than vision-based systems because of the lack of widespread acceptance of cameras due to privacy issues. In addition, low light or smoke can also hinder the recognition of vision, but sound waves are not affected by these special environmental situations.

"In the future, we will continue to study the complex activity and the situation of random positioning," Guo said. "As we know, human activities are complex and take the fall as an example.They can come in different postures.We hope to collect more falling activity data sets to achieve higher accuracy."

Guo said they will experiment with different numbers of sensors and their effectiveness in detecting and determining human activities. He said that there was an optimal number for the network that would make it viable for commercial and personal use in homes and buildings.