This research presents a detailed comparison of power consumption when deploying the YOLOv8 (You Only Look Once) object detection model on the NVIDIA Jetson Orin Nano, an edge AI device. The study examines two execution environments: a Docker containerized setup and a non-Docker (bare-metal) environment. Key metrics evaluated include the total wattage consumed during model operation, pre- and post-processing time, and the inference duration. A smart meter is used to capture real-time power usage data for both environments, ensuring accurate measurements. By comparing these parameters, the research sheds light on the impact of Docker containerization on power efficiency and processing performance. While Docker is widely used for its convenience and portability, it may introduce additional overheads that affect power consumption and execution speed, which are critical factors for AI applications deployed on edge devices with limited resources. The study seeks to determine if the trade-offs introduced by containerization are significant when compared to running YOLOv8 in a native environment. The results aim to provide actionable insights for developers and engineers looking to optimize power consumption and performance for AI models on edge devices.