Real-time institution video data analysis using fog computing and adaptive background subtraction

The increasing demand for video surveillance systems has led to a surge in research towards developing smart video surveillance systems that can combat the growing levels of insecurity. However, the massive amount of video data generated by these systems has overwhelmed the storage and processing capabilities of analytic applications. This paper proposes a fog computing-based smart video surveillance system that provides less latency, network bandwidth, and response time by localizing data to the edges of the network. The proposed system incorporates two key preprocessing steps, namely adaptive key frame extraction and adaptive contour-based background subtraction, to increase the quality of detecting abnormal motions from surveillance video streams. The adaptive key frame extraction mechanism extracts key frames from the available frames in the video sequence using a sliding window technique. The high-level semantic information of video frames is learned using the visual geometry group-16 Transfer Learning (VGG-16 TL) technique to better represent the video content. The adaptive contour-based background subtraction mechanism separates target foreground pixels from the background scenes, paving the way for easy detection of abnormal motions in the video frames. The proposed system leverages fog computing to process and store data, reducing latency and improving overall performance. The fog layer performs adaptive background subtraction, contour detection, and object analysis, ensuring timely processing of video frames and minimizing the need for transmitting large amounts of data to the cloud. The proposed system is evaluated using a real-time video dataset in terms of accuracy, compression ratio, precision, recall, processing time, latency, and network bandwidth, demonstrating the efficacy of the proposed system for abnormal motion detection and data transmission. Overall, the proposed fog computing-based smart video surveillance system provides an effective solution for detecting abnormal motions in real-time institution video data with reduced latency, network bandwidth, and response time, demonstrating the potential of fog computing for video surveillance applications.

Automated summary

This paper proposes a fog computing-based smart video surveillance system that provides less latency, network bandwidth, and response time by localizing data to the edges of the network. The proposed system incorporates adaptive key frame extraction and adaptive contour-based background subtraction, to increase the quality of detecting abnormal motions from surveillance video streams.