Real-Time Object Detection Using Raspberry Pi Zero 2W: An Optimized Approach

Abstract

Object detection's reliance on high computational density presents a major impediment to deployment on ubiquitous, resource-constrained edge systems. This research details an optimization regimen for the YOLOv5n single-stage detector, tailored for execution on embedded platforms. The Raspberry Pi Zero 2 W embodies the abstraction of computational power from physical scale, confining multi-core, 64-bit architecture (via the RP3A0 SiP) into a minimal footprint. Its creation conceptually validates the pursuit of democratized, high-density processing. This engineered pipeline achieves acceptable inference throughput on the constrained hardware, validating the methodology for realizing performant Edge Computer Vision applications in low power domains.