Abstract
ABSTRACT Checking the status of the power grid is very important. However, the low occurrence of defects in an actual power grid makes it difficult to collect training samples, which affects the training of defect-detection models. In this study, we proposed a method for enhancing the defective image of a power grid based on cycle-consistent adversarial networks (CycleGAN). The defective image sample dataset was expanded by fusing artificial defective samples, converted from defect-free components of samples with the trained CycleGAN model and updating its corresponding label file. Comparing the accuracy of the object detection model trained by the augmented dataset, we found a 2%–3% Average Precision (AP) improvement over baseline, and the fusing method of histogram specification reaches the best performance. In conclusion, the generative adversarial network (GAN) and its variants have considerable potential for dataset augmentation as well as scope for further improvement.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
