A Hybrid AI Pipeline for Real-Time Aerial Video Analytics on Resource-Limited Edge Devices with Performance Profiling
Received: 29 August 2025 | Revised: 16 September 2025 | Accepted: 24 September 2025 | Online: 16 October 2025
Corresponding author: Prashant V. Joshi
Abstract
The increasing demand for real-time aerial video analytics across diverse applications poses significant challenges to the deployment in resource-constrained edge devices. This study proposes a hybrid AI method that integrates computer vision preprocessing with deep learning inference in a pipelined architecture. The proposed method facilitates stage-wise processing, intelligent frame selection, and lightweight inference, thus enhancing efficiency in embedded edge devices. The method involves implementing and profiling multiple deep learning models—such as DenseNet-121, Inception v3, MobileNet v2, EfficientNet B2, and YOLO v8 (s, m, l)—on a Raspberry Pi 5 edge device. AI performance was evaluated using accuracy, mean Average Precision (mAP), and model size, while hardware profiling metrics (Cycles, Instructions per Cycle, Cache Refs, CPU Time, Memory, and Time Latency) were obtained through the perf profiler. The results show that EfficientNet B2 outperforms all other models, achieving the highest mAP (97.2%), the lowest energy consumption (0.98 J/Inf), low time latency (278 ms), and a compact model size (4.34 MB), making it highly suitable for real-time aerial video analytics in constrained edge environments. Overall, this study demonstrates a cost-effective and energy-efficient solution for real-time video analytics on edge devices.
Keywords:
AI, Unmanned Aerial Vehicles (UAVs), Tensorflow, YOLODownloads
References
Z. Cao, L. Kooistra, W. Wang, L. Guo, and J. Valente, ''Real-Time Object Detection Based on UAV Remote Sensing: A Systematic Literature Review,'' Drones, vol. 7, no. 10, Oct. 2023, Art. no. 620. DOI: https://doi.org/10.3390/drones7100620
A. Koubaa, A. Ammar, M. Alahdab, A. Kanhouch, and A. T. Azar, ''DeepBrain: Experimental Evaluation of Cloud-Based Computation Offloading and Edge Computing in the Internet-of-Drones for Deep Learning Applications,'' Sensors, vol. 20, no. 18, Jan. 2020, Art. no. 5240. DOI: https://doi.org/10.3390/s20185240
E. Badidi, K. Moumane, and F. E. Ghazi, ''Opportunities, Applications, and Challenges of Edge-AI Enabled Video Analytics in Smart Cities: A Systematic Review,'' IEEE Access, vol. 11, pp. 80543–80572, 2023. DOI: https://doi.org/10.1109/ACCESS.2023.3300658
R. Xu, S. Razavi, and R. Zheng, ''Edge Video Analytics: A Survey on Applications, Systems and Enabling Techniques,'' IEEE Communications Surveys & Tutorials, vol. 25, no. 4, pp. 2951–2982, 2023. DOI: https://doi.org/10.1109/COMST.2023.3323091
T. Bai, H. Zhao, L. Huang, Z. Wang, D. I. Kim, and A. Nallanathan, ''A Decade of Video Analytics at Edge: Training, Deployment, Orchestration, and Platforms,'' IEEE Communications Surveys & Tutorials, 2025. DOI: https://doi.org/10.1109/COMST.2025.3563377
V. Saikrishnan and M. Karthikeyan, ''Mayfly Optimization with Deep Learning-based Robust Object Detection and Classification on Surveillance Videos,'' Engineering, Technology & Applied Science Research, vol. 13, no. 5, pp. 11747–11752, Oct. 2023. DOI: https://doi.org/10.48084/etasr.6231
D. Ngo, H. C. Park, and B. Kang, ''Edge Intelligence: A Review of Deep Neural Network Inference in Resource-Limited Environments,'' Electronics, vol. 14, no. 12, Jan. 2025, Art. no. 2495. DOI: https://doi.org/10.3390/electronics14122495
Y. Wang, W. Wang, D. Liu, X. Jin, J. Jiang, and K. Chen, ''Enabling Edge-Cloud Video Analytics for Robotics Applications,'' IEEE Transactions on Cloud Computing, vol. 11, no. 2, pp. 1500–1513, Apr. 2023. DOI: https://doi.org/10.1109/TCC.2022.3142066
M. Bakirci, ''Performance evaluation of low-power and lightweight object detectors for real-time monitoring in resource-constrained drone systems,'' Engineering Applications of Artificial Intelligence, vol. 159, Nov. 2025, Art. no. 111775. DOI: https://doi.org/10.1016/j.engappai.2025.111775
R. M. A. Latif, K. Yoshigoe, N. Jamal, Y. Zhao, F. Ullah, and J. Ahmad, ''Edge-AI Architecture for Real-Time Computer Vision in Smart MediaApplications,'' in 2025 IEEE International Conference on Pattern Recognition, Machine Vision and Artificial Intelligence (PRMVAI), June 2025, pp. 1–6. DOI: https://doi.org/10.1109/PRMVAI65741.2025.11108373
C. Joshua et al., ''Latency-Accuracy Trade-off Analysis in Edge-Based Object Detection Pipelines," Advances in Engineering Software, vol. 213, pp. 1–13.
S. Kim, C. Kim, and S. Kim, ''Improving Performance of Real-Time Object Detection in Edge Device Through Concurrent Multi-Frame Processing,'' IEEE Access, vol. 13, pp. 1522–1533, 2025. DOI: https://doi.org/10.1109/ACCESS.2024.3520240
M. E. Isenkul, ''Energy-aware deep learning for real-time video analysis through pruning, quantization, and hardware optimization,'' Journal of Real-Time Image Processing, vol. 22, no. 3, June 2025, Art. no. 125. DOI: https://doi.org/10.1007/s11554-025-01703-0
P. V. Joshi, K. M. Sudharshan, M. G. Asuti, S. J. Poornashree, and S. K. Research, ''Real-Time Aerial Image Edge Detection with MicroWatt Power ISA on FPGA,'' in 2025 International Conference on Intelligent and Innovative Technologies in Computing, Electrical and Electronics (IITCEE), Jan. 2025, pp. 1–6. DOI: https://doi.org/10.1109/IITCEE64140.2025.10915495
"Annotated (Yolov8) Bridge Image for Detection." Kaggle, [Online]. Available: https://www.kaggle.com/datasets/sarahhdd/bridge.
"TDS-Satellite Images for Computer Vision Tasks." Kaggle, [Online]. Available: https://www.kaggle.com/datasets/dipensaini/sateelite-images-for-computer-vision-tasks.
K. P. Nandini and G. Seshikala, ''Efficient ECG Arrhythmia Detection on FPGA using Machine Learning and Fiducial Windowing,'' Engineering, Technology & Applied Science Research, vol. 15, no. 2, pp. 21100–21105, Apr. 2025. DOI: https://doi.org/10.48084/etasr.9589
J. Li and J. Ye, ''Edge-YOLO: Lightweight Infrared Object Detection Method Deployed on Edge Devices,'' Applied Sciences, vol. 13, no. 7, Jan. 2023, Art. no. 4402. DOI: https://doi.org/10.3390/app13074402
L. Rey et al., ''A Performance Analysis of You Only Look Once Models for Deployment on Constrained Computational Edge Devices in Drone Applications,'' Electronics, vol. 14, no. 3, Feb. 2025, Art. no. 638. DOI: https://doi.org/10.3390/electronics14030638
B. Huang, H. Lin, Z. Hu, X. Xiang, and J. Yao, ''An improved YOLOv3-tiny algorithm for vehicle detection in natural scenes,'' IET Cyber-Systems and Robotics, vol. 3, no. 3, pp. 256–264, 2021. DOI: https://doi.org/10.1049/csy2.12029
R. C. Camara de M. Santos, M. Coelho, and R. Oliveira, ''Real-time Object Detection Performance Analysis Using YOLOv7 on Edge Devices,'' IEEE Latin America Transactions, vol. 22, no. 10, pp. 799–805, July 2024. DOI: https://doi.org/10.1109/TLA.2024.10705971
Downloads
How to Cite
License
Copyright (c) 2025 S. J. Poornashree, Prashant V. Joshi, K. M. Sudharshan, Tony M. George
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain the copyright and grant the journal the right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) after its publication in ETASR with an acknowledgement of its initial publication in this journal.
