Remote Monitoring of Areas Using UAVs: An Approach Based on Transfer Learning and Federated Learning
Abstract
Remote terrain monitoring using unmanned aerial vehicles (UAVs) has gained prominence in fields such as precision agriculture, environmental mapping, and urban management, enabling large-scale and real-time data collection. This work presents a comparative analysis between centralized and federated machine learning, leveraging transfer learning for the classification of aerial images captured by UAVs. The MobileNetV3Small architecture was employed as the foundation for developing a model capable of maximizing computational efficiency on edge devices with processing and energy constraints, while the EuroSAT RGB dataset served as the evaluation basis. In the federated scenario, the DEEV (Devices, I Choose You) technique was employed to reduce the communication rate between devices and the central server, promoting greater bandwidth efficiency and data privacy. The results indicate that federated learning, combined with transfer learning, can achieve accuracy levels comparable to centralized learning, with significant advantages in terms of energy consumption and communication savings, demonstrating its feasibility for distributed monitoring applications.
References
Bengio, Y. (2012). Deep learning of representations for unsupervised and transfer learning. Volume 27, pages 17–36. PMLR.
Bichri, H., Chergui, A., and Mustapha, H. (2023). Image classification with transfer learning using a custom dataset: Comparative study. Procedia Computer Science, 220:48–54.
Dastour, H. and Hassan, Q. K. (2023). A comparison of deep transfer learning methods for land use and land cover classification. Sustainability, 15.
Gonçalves, A., Castro, A., Evilly, B., Leão, E., Neto, J. T., Silva, R., Filho, A. C., and Rabelo, R. (2024). Simulador de aplicações de inteligência artificial das coisas para monitoramento em tempo real. In Anais da XII Escola Regional de Computação do Ceará, Maranhão e Piauí, pages 159–168, Porto Alegre, RS, Brasil. SBC.
Hernandez-Sequeira, I., Fernandez-Beltran, R., and Pla, F. (2022). Transfer deep learning for remote sensing datasets: A comparison study. Pages 3207–3210.
Howard, A., Sandler, M., Chu, G., Chen, L.-C., Chen, B., Tan, M., Wang, W., Zhu, Y., Pang, R., Vasudevan, V., Le, Q. V., and Adam, H. (2019). Searching for MobileNetV3. CoRR, abs/1905.02244.
Hu, F., Xia, G.-S., Hu, J., and Zhang, L. (2015). Transferring deep convolutional neural networks for the scene classification of high-resolution remote sensing imagery. Remote Sensing, 7:14680–14707.
McMahan, B., Moore, E., Ramage, D., Hampson, S., and Arcas, B. A. y. (2017). Communication-efficient learning of deep networks from decentralized data. Volume 54, pages 1273–1282. PMLR.
Naushad, R., Kaur, T., and Ghaderpour, E. (2021). Deep transfer learning for land use and land cover classification: A comparative study. Sensors, 21.
Neto, J. R., Boukerche, A., Yokoyama, R. S., Guidoni, D. L., Meneguette, R. I., Ueyama, J., and Villas, L. A. (2017). Performance evaluation of unmanned aerial vehicles in automatic power meter readings. Ad Hoc Networks, 60:11–25.
Osco, L. P., Junior, J. M., Ramos, A. P. M., de Castro Jorge, L. A., Fatholahi, S. N., de Andrade Silva, J., Matsubara, E. T., Pistori, H., Gonçalves, W. N., and Li, J. (2021). A review on deep learning in UAV remote sensing. International Journal of Applied Earth Observation and Geoinformation, 102:102456.
Souza, A., Bittencourt, L., Cerqueira, E., Loureiro, A., and Villas, L. (2023). Dispositivos, eu escolho vocês: Seleção de clientes adaptativa para comunicação eficiente em aprendizado federado. Pages 1–14. SBC.
Yin, X., Chen, W., Wu, X., and Yue, H. (2017). Fine-tuning and visualization of convolutional neural networks. Pages 1310–1315.
Yosinski, J., Clune, J., Bengio, Y., and Lipson, H. (2014). How transferable are features in deep neural networks?
Zou, Q., Ni, L., Zhang, T., and Wang, Q. (2015). Deep learning based feature selection for remote sensing scene classification. IEEE Geoscience and Remote Sensing Letters, 12:2321–2325.
