Researchers benchmark object detection models for edge devices

Researchers have benchmarked several deep learning object detection models, including YOLOv8, EfficientDet Lite, and SSD variants, on various edge computing devices like Raspberry Pi and Jetson Orin Nano. The study evaluated performance based on energy consumption, inference time, and accuracy (mAP). Results indicate a trade-off between model accuracy and resource efficiency, with lower mAP models like SSD MobileNet V1 being faster and more energy-efficient, while higher mAP models like YOLOv8 Medium are more resource-intensive, though TPUs can mitigate this. The Jetson Orin Nano emerged as the most performant device for request handling. AI

IMPACT Provides guidance for optimizing deep learning model deployment on resource-constrained edge devices, balancing accuracy with efficiency.

RANK_REASON This is a research paper evaluating existing models and hardware for a specific application.

Read on arXiv cs.LG →

• Daghash K. Alqahtani
• EfficientDet Lite
• Jetson Orin Nano
• Raspberry Pi
• SSD MobileNet V1
• YOLOv8
• YOLOv8 Medium

• paper
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →