Efficient Person Detection on Single Board Computers: A Comparative Analysis of Algorithms

Abstract

This research paper conducts an exhaustive technical evaluation and comparative analysis of person detection algorithms optimized explicitly for deployment on single-board computers (SBCs). The burgeoning demand for edge computing solutions, particularly on constrained platforms like SBCs, necessitates the optimization of object detection algorithms, specifically targeting person detection. This study meticulously examines and contrasts the performance of varied person detection algorithms tailored to SBCs, meticulously scrutinizing metrics encompassing accuracy, computational efficiency, CPU utilization, and latency. By conducting rigorous experimentation and in-depth analysis, this research dissects the intricate trade-offs inherent in algorithmic intricacy, computational efficacy, and the real-time inference capabilities crucial for edge computing scenarios. The research outcomes yield profound technical insights essential for strategic algorithm selection, aiding in the identification of the most adept person detection algorithm primed for deployment on SBCs across diverse applications, spanning surveillance systems to edge-driven robotics. Ultimately, this comprehensive comparative analysis serves as a robust technical blueprint, offering vital guidance for refining person detection algorithms on resource-constrained edge devices, thereby propelling advancements in edge computing and bolstering real-time object detection capabilities.