Multi-sensors Fused IoT-based Home Surveillance via Bag of Visual and Motion Features

Internet of things (IoT) represent the small devices connected together wirelessly collecting data to make lifestyle convenient. Inertial measurement units (IMU) and cameras connected together to collect data from multiple indoor activities can also support home surveillance systems. The traditional closed-circuit television is out-fashioned due to the huge volume of storage requirements and not connected together to notify users immediately of apprehensive activities. Therefore, this paper proposes an IoT-based surveillance system for indoor environments that will upkeep the security methods inside the home. For this purpose, the fused multisensors-based data is acquired from two state-of-the-art datasets, namely, opportunity++ and CMU-MMAC. This acquired data from IoT devices is further pre-processed through multiple filtering techniques according to the type of data. Then, a skeleton model has been designed for the filtered video frame sequences. Furthermore, a bag of visual and motion features has been extracted using three different techniques followed by their discrimination. Finally, the IoT-based surveillance system detects indoor activities and provides feedback to the user.