A unified smart surveillance system incorporating adaptive foreground extraction and deep learning-based classification

The smart surveillance system is one of the most important services provided in a smart city. The smart surveillance applications are equipped with camera sensors to automatically detect and identify potential regions through automated object detection methods. Usually, such methods require high-complexity image processing techniques and algorithms. Hence, the design of low-complexity automated object detection algorithms becomes an important topic in this area. We consider a foreground extraction and deep learning to achieve these goals. A novel unified technique is proposed to detect a moving object from the surveillance videos based on CPU (central processing units). Unlike most of the existing methods that are relying on pixel information, we use a block-based texture and spatial-temporal information. We use this method to determine the area of the detected moving object(s) only. Furthermore, the area will be processed through a deep learning-based image classification in GPU (graphics processing units) in order to ensure high efficiency and accuracy. It cannot only help to detect object rapidly and accurately, but also can reduce big data volume needed to be stored in smart surveillance systems.