Resilient system for multi-tracking problem using several cameras on Raspberry Pi

Video surveillance and monitoring is an active field of research nowadays. These systems integrate sensors and algorithms which extract features from a data stream enabling analysis. In this paper, after studying some existing video-monitoring systems, we take the best from characteristics to compensate the limitations of video surveillance system as occlusions and sensors failure. The objective of this thesis is the implementation of a system performing a continuous tracking in case of these limitations. Our architecture is agent-based and inspired by an organizational and holonic multi-agent model. It paves the way to scalable, modular, resilient and open multi-sensor and multi-method surveillance systems. Finally, the system is validated through a case study representing vehicles tracked across multiple cameras.