With vigorous development e.g., in autonomous driving and remote sensing, oriented object detection has gradually been featured. The majority of existing methods directly perform regression on the rotation angle, whic...
详细信息
With vigorous development e.g., in autonomous driving and remote sensing, oriented object detection has gradually been featured. The majority of existing methods directly perform regression on the rotation angle, which we argue has fundamental limitations of boundary discontinuity (even if using Gaussian or RotatedIoU-based losses). In this paper, a novel angle coder named phase-shifting coder (PSC) is proposed to address this issue. Different from another well-explored alternative i.e., angle classification, PSC achieves boundary-discontinuity-free in a continuous and differentiable manner and thus can work together with Gaussian or RotatedIoU-based methods to further boost their performance. Moreover, by rethinking the boundary discontinuity of elongated and square-like objects as rotational symmetry of different cycles, a dual-frequency version (PSCD) is proposed to accurately predict the orientation of both types of objects. Visual analysis and extensive experiments on several popular backbone detectors and datasets demonstrate the effectiveness and the potentiality of our approach. When facing scenarios requiring high-quality bounding boxes, the proposed methods are expected to give a competitive performance.
Wind turbine blades (WTBs) are critical components of wind turbines. Exposed directly to harsh environmental conditions, they are susceptible to various defects. Timely detection of surface defects on WTBs is crucial ...
详细信息
Wind turbine blades (WTBs) are critical components of wind turbines. Exposed directly to harsh environmental conditions, they are susceptible to various defects. Timely detection of surface defects on WTBs is crucial for effective repair. Utilizing unmanned aerial vehicles (UAVs) for wind farm inspection offers greater efficiency and enhanced safety compared to manual inspections. However, the variability in angle and distance among images collected by UAVs results in significant variations in defect size and shape, including tilted targets and small targets. To address these issues, we propose a method for detecting surface defects on WTBs based on improved YOLOv7. To better characterize tilted defects, the angle prediction branch based on the phase-shifting coder (PSC) is integrated into the YOLOv7 to enable rotation of the target detection box. Additionally, efficient multi-scale attention (EMA) is adapted to prioritize defective parts and improve model precision. Finally, a normalized Wasserstein distance (NWD) loss function is introduced to mitigate the sensitivity to positional deviations and improve the detection precision for small defects. The experimental results show that the mean average precision (mAP) of the proposed algorithm reaches 89.1%, surpassing the original YOLOv7 algorithm and other rotating target detection algorithms.
暂无评论