Application of individual activity recognition in the room using CNN Alexnet method

Abstract

Abstract Technological developments in the digital age are experiencing rapid improvement, especially in the field of sensing technology and network infrastructure. Where due to this, the technology that is more sophisticated and low prices can be affordable by ordinary people. Developments in the field of sensing technology and network infrastructure have enabled the development of intelligent software that can provide real-time analysis of certain situations in an environment, with the aim of improving the quality of human life [1]. Security camera systems, such as Closed-circuit Television (CCTV) are widely used, and can be found in places where monitoring is needed [2]. Low prices and increased use of CCTV in many areas make it easy to use. However, this traditional CCTV requires humans to monitor scenes continuously [3]. Of course, it is very inefficient to ask people to monitor the scene. Therefore, we need an individual activity detection application in the room that is able to recognition to human activities. To detect human activity, aspect ratio and Euclidean Distance are used, while for the recognition, the Alexnet Architecture Convolutional Neural Network method is used. The test results obtained a success rate of 100% for detection of human activity, a best success rate of 96% for recognizing human activity.