Design and IoT Implementation of a LiDAR Assisted Large-Area Video Surveillance Network

Over the years, the advancement of technology has evolved the surveillance system to a new dimension in comparison with the age-old approach, which is highly human-dependent from multiple aspects, such as patrolling, tracking a suspect, monitoring, and so on. As the modernization of the surveillance system makes it more sophisticated, on the downside, it imposes several technical challenges, like maintenance cost, scalability of the network, and others. In this paper, we have tried to address a few of those challenges to make the surveillance system more efficient. The proposed design of a multi-sensor data fusion, along with the transfer learning and the implementation of the Long Range (LoRa) technology based Internet-of-Things (IoT) framework, provides an efficient system with some degree of scalability. As the multiple sensors are utilized, the storage consumption has reduced significantly along with the computational cost. In addition to that, transfer learning method based video analysis has enhanced its efficiency, which reduces the human effort for monitoring. Lastly, the LoRa based IoT framework allows us to cover a large geographical area with a provision of scaling up the whole architecture as per the requirement. Nevertheless, the overall performance of this proposed framework has been evaluated, and it yields highly accurate performance in different scenarios.