Optimizing Wind Turbine Surface Defect Detection: A Rotated Bounding Box Approach

This repository contains rotated bounding box annotations for DTU-Drone inspection images and Blade30 datasets. The names of the original images are changed to image1, 2 and 3 for simplicity and some synthetic images are also generated for underrepresented classes to balance the dataset (only for DTU dataset).

Introduction

Detecting surface defects on Wind Turbine Blades (WTBs) from remotely sensed images is a crucial step toward automated visual inspection. Typical object detection algorithms use standard bounding boxes to locate defects on WTBs. However, Oriented Bounding Boxes (OBBs) have been shown in cases of satellite imagery, to provide more precise localization of object regions and actual orientation. Existing WTB datasets do not depict defects using OBBs and this causes the lack of useful orientational information. In this paper, we consider OBBs for WTB surface defect detection through two publicly available datasets, introducing new annotations to the community. Baselines were constructed on state-of-the-art rotated object detectors, demonstrating considerable promise and known gaps that can be addressed in the future. We present a comprehensive analysis of their performances including ablation study and discussions on the importance of angular disparity between OBBs.

Standard Bounding Box vs. Oriented Bounding Box:

.

Advantages Over standard BBox

• OBB localization is more tighter than standard BB
• Help in reducing overlapping areas
• Better for dense objects

Visual Results

Comapring Jensen-Shannon Distribution (JSD) - symmetric, Gaussian Wasserstein Distance (GWD), RIoU and Hybrid Loss function. .

Project Poster

Datasets

DTU-Drone Inspection images and Blade30

Citation

@INPROCEEDINGS{10715208,
  author={Gohar, Imad and Halimi, Abderrahim and Yew, Weng Kean and See, John},
  booktitle={2024 32nd European Signal Processing Conference (EUSIPCO)}, 
  title={Optimizing Wind Turbine Surface Defect Detection: A Rotated Bounding Box Approach}, 
  year={2024},
  volume={},
  number={},
  pages={631-635},
  keywords={Location awareness;Visualization;Soft sensors;Signal processing;Wind turbines;Satellite images;Reliability;Remote sensing;Standards;Defect detection;wind turbine blade;surface defect detection;oriented bounding boxes;UAV remote sensing},
  doi={10.23919/EUSIPCO63174.2024.10715208}}
}