Performance characterization on embedded systems for Edge AI person-detection models
Article Sidebar
Main Article Content
Abstract
This paper presents a hardware performance characterization for two Edge AI platforms: Raspberry Pi 4 and NVIDIA Jetson Nano, for the task of automatic people detection using a deep learning model. For comparison purposes, we use the MLPerf Inference Benchmark evaluation system. The characterization considers the results from an SSD-Mobilenet object-detection model using two different datasets, one with 80 different object classes and another with only people. Comparison metrics consider model accuracy, latency, queries processed per second, and samples processed per second under the evaluation of different execution scenarios.
Article Details
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Los autores conservan los derechos de autor y ceden a la revista el derecho de la primera publicación y pueda editarlo, reproducirlo, distribuirlo, exhibirlo y comunicarlo en el país y en el extranjero mediante medios impresos y electrónicos. Asimismo, asumen el compromiso sobre cualquier litigio o reclamación relacionada con derechos de propiedad intelectual, exonerando de responsabilidad a la Editorial Tecnológica de Costa Rica. Además, se establece que los autores pueden realizar otros acuerdos contractuales independientes y adicionales para la distribución no exclusiva de la versión del artículo publicado en esta revista (p. ej., incluirlo en un repositorio institucional o publicarlo en un libro) siempre que indiquen claramente que el trabajo se publicó por primera vez en esta revista.
References
[1] B. Varghese, N. Wang, D. Bermbach, C.-H. Hong, E. D. Lara, W. Shi y C. Stewart, «A Survey on Edge Performance Benchmarking,» ACM Computing Surveys, vol. 54, nº 3, pp. 1-33, Abril 2022.
[2] Z. Zhou, K. Chen, Z. Shi, Y. Guo y J. Ye, «Object Detection in 20 Years: A Survey,» Proceedings of the IEEE, vol. 111, nº 3, pp. 257-276, Marzo 2023.
[3] V. J. Reddi, C. Cheng, D. Kanter, P. Mattson, G. Schmuelling, C.J. Wu, B. Anderson, M. Breughe, M. Charlebois, W. Chou et al., «MLPerf Inference Benchmark,» ACM/IEEE 47th Annual International Symposium on Computer Architecture (ISCA), pp. 446-459, 2020.
[4] E. Upton y G. Halfacree, Raspberry Pi user guide, John Wiley & Sons, 2016.
[5] F. N. Uzun, M. Kayrici y B. Akkuzu, «Nvidia Jetson Nano Development Kit,» Programmable Smart Microcontroller Cards, p. 82, 2021.
[6] W. Liu, D. Anguelov, D. Erhan, C. Szegedy, S. Reed, C.-Y. Fu y A. C. Berg, «SSD: Single Shot MultiBox Detector,» de European conference on computer vision, 2016.
[7] A. G. Howard, M. Zhu, B. Chen, D. Kalenichenko, W. Wang, T. Weyand, M. Andreetto y H. Adam, «MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications,» Publisher: arXiv Version Number: 1, 2017. [En línea]. Available: https://arxiv.org/abs/1704.04861.
[8] T.-Y. Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Dollár y C. L. Zitnick, «Microsoft COCO: Common Objects in Context,» Computer Vision – ECCV 2014, vol. 8693, pp. 740-755, 2014.
[9] «COCO dataset - Common Objects in Context,» [En línea]. Available: https://cocodataset.org/#home. [Último acceso: Julio 2025].
[10] R. Padilla, S. L. Netto y E. Da Silva, «A survey on performance metrics for object-detection algorithms,» de International conference on systems, signals and image processing (IWSSIP, 2020.
[11] W. Wang, W. Hong, F. Wang y J. Yu, «GAN-Knowledge Distillation for One-Stage Object Detection,» IEEE Access, vol. 8, pp. 60719--60727, 2020.
[12] Y. Hu, N. Chen, Y. Hou, X. Lin, B. Jing y P. Liu, «Lightweight deep learning for real-time road distress detection on mobile devices,» Nature Communications, vol. 16, nº 1, 2025.
[13] NVIDIA Developer, «Documentation Jetson Nano 2GB Developer Kit,» [En línea]. Available: https://developer.nvidia.com/embedded/learn/jetson-nano-2gb-devkit-user-guide. [Último acceso: Julio 2025].
[14] Y.-H. Chang, J. Pu, W.-m. Hwu y J. Xiong, «MLHarness: A scalable benchmarking system for MLCommons,» BenchCouncil Transactions on Benchmarks, Standards and Evaluations, vol. 1, 2021.