Concept For Using Fpga-Based Computer Vision In Smart-Home IoT Paradigm

Leonid Poskotin

doi:10.63313/AJET.9047

Authors

Leonid Poskotin School of Electronic Engineering, Tianjin University of Technology and Education, Tianjin 300222, China Author

DOI:

https://doi.org/10.63313/AJET.9047

Keywords:

Internet of Things, Field-programmable Gate Array, Computer Vision

Abstract

The smart home IoT sector continues to grow every year, with an ever-increasing number of new solutions emerging that incorporate computer vision technologies using CNN-based machine learning approaches. For the most part, CNN computations are performed either on a cloud server—which can increase computation time to 100 milliseconds and leave the data vulnerable to uncontrolled cyberattacks—or on edge devices powered by expensive GPUs or low-performance CPUs. This article describes a concept for an FPGA-based device built on the ZYNQ platform, which, in theory, could resolve these issues. The proposed concept is capable of capturing video feed using an OV5640 camera, processing it, and performing object detection using a CNN model. The system can manage and control most types of IoT devices using the Home Assistant software, installed as a Docker container on a custom-built Linux distribution. As part of this paper, an experiment was conducted to compare the inference times of an FPGA-based and a CPU-based CNN accelerator on YOLOv3 model. According to the results of the experiment, the FPGA-based solution performed best. This result could serve as a foundation for future work on integrating FPGAs into the IoT paradigm.

References

[1] Maslej, N., Fattorini, L., Perrault, R., Gil, Y., Parli, V., Kariuki. N., Capstick, E., Reuel, A., Brynjolfsson, E., Etchemendy, E., Ligett, K., Lyons, T., Manyika, J., Niebles, J.C., Shoham, Y., Wald, R., Walsh, T., Hamrah, A., Santarlasci, L., Lotufo, B.L., Rome, A., Shi, A., Oak, S. (2025) The AI Index 2025 Annual Report. AI Index Steering Committee, Institute for Human-Centered AI, Stanford University, Stanford, CA, April 2025.

https://doi.org/10.48550/arXiv.2504.07139

[2] Elouardi, S., Motii, A., Jouhari, M., Nasser Hassane Amadou, A. and Hedabou, M. (2024) A Survey on Hybrid-CNN and LLMs for Intrusion Detection Systems: Recent IoT Datasets. IEEE Access, 12, 180009-180033.

https://doi.org/10.1109/ACCESS.2024.3506604

[3] Khan, H., Yuan, X., Qingge, L. and Roy, K. (2025) Violence Detection From Industrial Surveillance Videos Using Deep Learning. IEEE Access, 13, 15363-15375, 2025.

https://doi.org/10.1109/ACCESS.2025.3531213.

[4] Aldossary, M., Alharbi, H.A. and Anwar Ul Hassan, C. (2024) Internet of Things (IoT)-Enabled Machine Learning Models for Efficient Monitoring of Smart Agriculture. IEEE Access, 12, 75718-75734.

https://doi.org/10.1109/ACCESS.2024.3404651

[5] Gad MM, Gad W, Abdelkader T, Naik K. (2025) Personalized Smart Home Automation Using Machine Learning: Predicting User Activities. Sensors (Basel). 2025 Oct 2;25(19),6082.

https://doi.org/10.3390/s25196082

[6] Wang, F., Zhang, M., Wang, X., Ma, X. and Liu, J. (2020) Deep Learning for Edge Computing Applications: A State-of-the-Art Survey. IEEE Access, 8, 58322-58336, 2020.

https://doi.org/10.1109/ACCESS.2020.2982411

[7] Rosero-Montalvo, P.D., Tözün, P. and Hernandez, W. (2024) Optimized CNN Architectures Benchmarking in Hardware-Constrained Edge Devices in IoT Environments. IEEE Internet of Things Journal, 11, 20357-20366, 1 June1, 2024.

https://doi.org/10.1109/JIOT.2024.3369607