Avaliação Multicritério de Detectores Leves de Pessoas para Edge Computing em SBCs: Desempenho, Latência e Eficiência Energética
Resumo
Este trabalho apresenta uma avaliação comparativa de detectores leves de pessoas para computação de borda nas plataformas Jetson Nano e Raspberry Pi 4B. Foram avaliados os modelos YOLOv3-tiny, YOLOv5n, YOLOv7tiny, YOLOv8n e SSDLite320 MobileNetV3-Large sob um protocolo padronizado, utilizando ONNX Runtime em CPU. A análise considerou desempenho preditivo, latência, FPS, uso de memória, temperatura e consumo energético. Os resultados dos experimentos indicam que os modelos da família YOLO obtiveram melhor desempenho preditivo geral, enquanto o YOLOv5n apresentou o melhor equilíbrio entre qualidade de detecção, latência e custo energético. Além disso, a Raspberry Pi 4B mostrou maior eficiência energética em todos os modelos avaliados.
Referências
Ali, M. L. and Zhang, Z. (2024). The yolo framework: A comprehensive review of evolution, applications, and benchmarks in object detection. Computers, 13(12).
Alqahtani, D. K., Cheema, M. A., and Toosi, A. N. (2025). Benchmarking deep learning models for object detection on edge computing devices. In Gaaloul, W., Sheng, M., Yu, Q., and Yangui, S., editors, Service-Oriented Computing, pages 142–150, Singapore. Springer Nature Singapore.
Alqahtani, D. K., Rodriguez, M. A., Cheema, M. A., Rezatofighi, H., and Toosi, A. N. (2026). Ecore: Energy-conscious optimized routing for deep learning models at the edge.
Azab, E., Ehab, M., Shihata, L., and Mashaly, M. (2026). Optimizing computer vision for edge deployment in industry 4.0: A framework and experimental evaluation. Technologies, 14(2):126.
Cocks, G., Magbag, T., Hemmati, M., and Wang, K. I.-K. (2023). Real-time pedestrian detection using resource constrained embedded platforms – a review. In 2023 IEEE Smart World Congress (SWC), pages 1–8.
Fan, H., Liu, S., Ferianc, M., Ng, H.-C., Que, Z., Liu, S., Niu, X., and Luk, W. (2018). A real-time object detection accelerator with compressed ssdlite on fpga. In 2018 International Conference on Field-Programmable Technology (FPT), pages 14–21.
Hemmatirad, K., Babaie, M., Afshari, M., Maleki, D., Saiadi, M., and Tizhoosh, H. R. (2022). Quality control of whole slide images using the yolo concept. In 2022 IEEE 10th International Conference on Healthcare Informatics (ICHI), pages 282–287.
Howard, A., Sandler, M., Chen, B., Wang, W., Chen, L.-C., Tan, M., Chu, G., Vasudevan, V., Zhu, Y., Pang, R., Adam, H., and Le, Q. (2019). Searching for mobilenetv3. In 2019 IEEE/CVF International Conference on Computer Vision (ICCV), pages 1314–1324.
Hussain, M. (2024). Yolov1 to v8: Unveiling each variant–a comprehensive review of yolo. IEEE Access, 12:42816–42833.
Islam, R. B., Akhter, S., Iqbal, F., Saif Ur Rahman, M., and Khan, R. (2023). Deep learning based object detection and surrounding environment description for visually impaired people. Heliyon, 9(6).
Jain, S., Kumar, A., Goswami, P., Khajuria, R., Dayal, P., and Banerjee, S. (2024). Efficient person detection on single board computers: A comparative analysis of algorithms. In 2024 10th International Conference on Advanced Computing and Communication Systems (ICACCS). IEEE.
Jocher, G. and Qiu, J. (2026). Ultralytics yolo26. [link]. Version 26.0.0. Software. Licença AGPL-3.0. Acesso em: 24/04/2026.
Kim, C. E., Oghaz, M. M. D., Fajtl, J., Argyriou, V., and Remagnino, P. (2019). A comparison of embedded deep learning methods for person detection. In Proceedings of the 14th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2019), pages 459–465. SciTePress.
Kolosov, D., Kelefouras, V., Kourtessis, P., and Mporas, I. (2022). Anatomy of deep learning image classification and object detection on commercial edge devices: A case study on face mask detection. IEEE Access, 10:115313–115333.
Lazarevich, I., Grimaldi, M., Kumar, R., Mitra, S., Khan, S., and Sah, S. (2023). Yolo-bench: Benchmarking efficient object detectors on embedded systems. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, pages 1169–1178.
Redmon, J., Divvala, S., Girshick, R., and Farhadi, A. (2016). You only look once: Unified, real-time object detection. In 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 779–788.
Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., and Chen, L.-C. (2018). Mobilenetv2: Inverted residuals and linear bottlenecks. In 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4510–4520.
Terven, J., Córdova-Esparza, D.-M., and Romero-González, J.-A. (2023). A comprehensive review of yolo architectures in computer vision: From yolov1 to yolov8 and yolonas. Machine Learning and Knowledge Extraction, 5(4):1680–1716.
Ultralytics (2024a). Ultralytics yolo11. [link]. Acesso em: 2026-03-11.
Ultralytics (2024b). Yolov9: A leap forward in object detection technology. [link]. Acesso em: 2026-03-11.
Wang, A., Chen, H., Liu, L., Chen, K., Lin, Z., Han, J., and Ding, G. (2024). Yolov10: Real-time end-to-end object detection. arXiv preprint arXiv:2405.14458.
Yilmaz, E. N. and Navruz, T. S. (2025). Real-time object detection: A comparative analysis of yolo, ssd, and efficientdet algorithms. In 2025 7th International Congress on Human-Computer Interaction, Optimization and Robotic Applications (ICHORA), pages 1–9.
Alqahtani, D. K., Cheema, M. A., and Toosi, A. N. (2025). Benchmarking deep learning models for object detection on edge computing devices. In Gaaloul, W., Sheng, M., Yu, Q., and Yangui, S., editors, Service-Oriented Computing, pages 142–150, Singapore. Springer Nature Singapore.
Alqahtani, D. K., Rodriguez, M. A., Cheema, M. A., Rezatofighi, H., and Toosi, A. N. (2026). Ecore: Energy-conscious optimized routing for deep learning models at the edge.
Azab, E., Ehab, M., Shihata, L., and Mashaly, M. (2026). Optimizing computer vision for edge deployment in industry 4.0: A framework and experimental evaluation. Technologies, 14(2):126.
Cocks, G., Magbag, T., Hemmati, M., and Wang, K. I.-K. (2023). Real-time pedestrian detection using resource constrained embedded platforms – a review. In 2023 IEEE Smart World Congress (SWC), pages 1–8.
Fan, H., Liu, S., Ferianc, M., Ng, H.-C., Que, Z., Liu, S., Niu, X., and Luk, W. (2018). A real-time object detection accelerator with compressed ssdlite on fpga. In 2018 International Conference on Field-Programmable Technology (FPT), pages 14–21.
Hemmatirad, K., Babaie, M., Afshari, M., Maleki, D., Saiadi, M., and Tizhoosh, H. R. (2022). Quality control of whole slide images using the yolo concept. In 2022 IEEE 10th International Conference on Healthcare Informatics (ICHI), pages 282–287.
Howard, A., Sandler, M., Chen, B., Wang, W., Chen, L.-C., Tan, M., Chu, G., Vasudevan, V., Zhu, Y., Pang, R., Adam, H., and Le, Q. (2019). Searching for mobilenetv3. In 2019 IEEE/CVF International Conference on Computer Vision (ICCV), pages 1314–1324.
Hussain, M. (2024). Yolov1 to v8: Unveiling each variant–a comprehensive review of yolo. IEEE Access, 12:42816–42833.
Islam, R. B., Akhter, S., Iqbal, F., Saif Ur Rahman, M., and Khan, R. (2023). Deep learning based object detection and surrounding environment description for visually impaired people. Heliyon, 9(6).
Jain, S., Kumar, A., Goswami, P., Khajuria, R., Dayal, P., and Banerjee, S. (2024). Efficient person detection on single board computers: A comparative analysis of algorithms. In 2024 10th International Conference on Advanced Computing and Communication Systems (ICACCS). IEEE.
Jocher, G. and Qiu, J. (2026). Ultralytics yolo26. [link]. Version 26.0.0. Software. Licença AGPL-3.0. Acesso em: 24/04/2026.
Kim, C. E., Oghaz, M. M. D., Fajtl, J., Argyriou, V., and Remagnino, P. (2019). A comparison of embedded deep learning methods for person detection. In Proceedings of the 14th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2019), pages 459–465. SciTePress.
Kolosov, D., Kelefouras, V., Kourtessis, P., and Mporas, I. (2022). Anatomy of deep learning image classification and object detection on commercial edge devices: A case study on face mask detection. IEEE Access, 10:115313–115333.
Lazarevich, I., Grimaldi, M., Kumar, R., Mitra, S., Khan, S., and Sah, S. (2023). Yolo-bench: Benchmarking efficient object detectors on embedded systems. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, pages 1169–1178.
Redmon, J., Divvala, S., Girshick, R., and Farhadi, A. (2016). You only look once: Unified, real-time object detection. In 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 779–788.
Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., and Chen, L.-C. (2018). Mobilenetv2: Inverted residuals and linear bottlenecks. In 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4510–4520.
Terven, J., Córdova-Esparza, D.-M., and Romero-González, J.-A. (2023). A comprehensive review of yolo architectures in computer vision: From yolov1 to yolov8 and yolonas. Machine Learning and Knowledge Extraction, 5(4):1680–1716.
Ultralytics (2024a). Ultralytics yolo11. [link]. Acesso em: 2026-03-11.
Ultralytics (2024b). Yolov9: A leap forward in object detection technology. [link]. Acesso em: 2026-03-11.
Wang, A., Chen, H., Liu, L., Chen, K., Lin, Z., Han, J., and Ding, G. (2024). Yolov10: Real-time end-to-end object detection. arXiv preprint arXiv:2405.14458.
Yilmaz, E. N. and Navruz, T. S. (2025). Real-time object detection: A comparative analysis of yolo, ssd, and efficientdet algorithms. In 2025 7th International Congress on Human-Computer Interaction, Optimization and Robotic Applications (ICHORA), pages 1–9.
Publicado
25/05/2026
Como Citar
CARDOZO, Daniel Morais; BRUSCHI, Sarita Mazzini.
Avaliação Multicritério de Detectores Leves de Pessoas para Edge Computing em SBCs: Desempenho, Latência e Eficiência Energética. In: WORKSHOP DE INTELIGÊNCIA ARTIFICIAL PARA REDES DE COMPUTADORES (WIARC), 1. , 2026, Praia do Forte/BA.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 43-56.
DOI: https://doi.org/10.5753/wiarc.2026.22903.
