Multi-Camera People Counting Using a Queue-Buffer Algorithm for Effective Search and Rescue in Building Disasters

Over the past decade, the frequency of building disasters has been concerning. However, search and rescue operations often encounter challenges due to limited information, resulting in delays or inadequate assistance for occupants trapped inside the building. The importance of obtaining real-time occupancy information has been recognized and studied in the field of people counting. While various sensor studies have been proposed, using camera sensors to accurately count occupants in a building has been attempted. Nevertheless, even with highly accurate camera sensors, counting errors (i.e., miscounting) are unavoidable due to factors such as suboptimal accuracy and occlusion issues in single-camera setups as well as the miscounts occurred by the blind spots between multiple cameras. To overcome these limitations, this paper introduces a novel people counting method called the "Queue-Buffer Algorithm", a multi-camera-based error-correction algorithm for real-time people counting that leverages contextual data to correct miscounts. By implementing this multi-camera people counting approach, the accuracy and efficiency of search and rescue operations can be significantly improved, leading to enhanced outcomes in terms of lives saved and timely evacuation.