Gunawan, Teddy Surya and Mazlan, Muhammad Harith Aiman and Kartiwi, Mira and Md Yusoff, Nelidya (2025) Robust drone detection in adverse weather using YOLOv11 with synthetic rain augmentation. In: 11th IEEE International Conference on Smart Instrumentation, Measurement and Applications, ICSIMA 2025, 10-11 September 2025, Kuala Lumpur, Malaysia.
|
PDF
- Published Version
Restricted to Registered users only Download (1MB) | Request a copy |
||
|
PDF
- Supplemental Material
Download (129kB) | Preview |
Abstract
This paper investigates the performance of YOLOv11 models for drone detection under adverse weather simulated via synthetic rain at five intensity levels. Five YOLOv11 variants, n, s, m, l, and x, were fine-tuned using rain-augmented datasets. Results show a consistent degradation in accuracy with increasing rain intensity. Among the models, YOLOv11m achieved the best trade-off, maintaining mAP@0.5 above 0.92 and F1-scores above 0.88 under clear to moderate rain. Fine-tuning significantly improved recall and fitness scores compared to baseline performance. A representative confusion matrix yielded 310 true positives, 59 false negatives, and 37 false positives, illustrating the inherent precision–recall trade-off under visual degradation. Compared to earlier YOLO versions, YOLOv11 demonstrated improved resilience to synthetic rain effects such as occlusion and blur. However, performance dropped sharply under heavy rain conditions. These findings support the application of rain-augmented fine-tuning for robust drone detection. Future work includes sensor fusion and real-world weather validation.
Actions (login required)
![]() |
View Item |

Download Statistics
Download Statistics