Wind turbine defect detection, drone imaging, few-shot detection.
Enhanced Drone Surveillance & Wind Turbine Defect Detection
End-to-end pipeline for automated wind turbine blade inspection. UAVs capture high-resolution aerial imagery of wind turbine farms. Key challenges include data scarcity, the need for substantial computational resources, and the geometric complexities in accurately localizing defects.
End-to-end UAV-based wind turbine blade inspection pipeline
Slice-Aided Defect Detection in Ultra High-Resolution Images
This work addresses the processing challenges of high-resolution drone imagery with small objects. The framework compares different slicing strategies for detecting surface damage on turbine blades, enabling accurate localization of small defects in ultra high-resolution images.
Slice-aided defect detection on turbine blade surfaces
Few-Shot Detection of Wind Turbine Blade Surface Defects
Two-stage few-shot object detection framework. This approach tackles severe class imbalance and limited annotated data using few-shot object detection methodology with Contrastive Proposal Encoding loss, achieving up to 22% improvement in novel class detection.
Few-shot object detection for blade surface defects
- I. Gohar, A. Halimi, W. K. Yew and J. See, "Review of state-of-the-art surface defect detection on wind turbine blades through aerial imagery: Challenges and recommendations," Engineering Applications of AI, 2025.
- I. Gohar, A. Halimi, J. See, W. K. Yew, C. Yang, "Slice-Aided Defect Detection in Ultra High-Resolution Wind Turbine Blade Images," Machines, 2023.
- I. Gohar, A. Halimi, W. K. Yew and J. See, "Addressing Class Scarcity and Imbalance for Few-Shot Detection of Wind Turbine Blade Surface Defects," ISPACS, Indonesia, November 2025. Best Paper Award.