Attention-Guided Autoencoder Fusion for Insulator Defect Detection Using UAV Transmission-Line Imaging
Researchers have developed a new framework called AE-YOLO for detecting defects in high-voltage transmission-line insulators using drone imagery. This system integrates autoencoders and attention mechanisms to improve feature discrimination and handle challenges like class imbalance and scale variation. The framework combines multiple YOLO models and uses a novel regularization strategy to enhance detection accuracy, achieving state-of-the-art results on a specialized dataset. AI
IMPACT This framework offers a more accurate and scalable solution for critical infrastructure inspection, potentially improving safety and efficiency in power grid maintenance.