Transmission tower detection is a crucial step in smart power survey and inspection

especially in highaltitude areas with complex terrain and harsh environments

where automated detection faces severe challenges.Satellite remote sensing data is easily accessible and publicly labeled

but its resolution is relatively low.UAV remote sensing technology can efficiently obtain high-resolution Earth observation data

but the training of intelligent detection models in complex scenes relies heavily on manual data annotation

which is time-consuming and laborious.To address these issues

this paper proposes an unsupervised generative cross-domain transmission tower detection method from satellite to UAV.Firstly

it analyzes the reasons why the differences between satellite and UAV data lead to difficulties in the generalization of deep learning models.Then

a two-step alignment strategy is designed

which achieves scale alignment through observation window normalization and utilizes a generative model for image style transfer to reduce the differences between the two.Finally

labeled satellite data is converted into UAV-like data to construct a training set

enabling unsupervised detection of UAV observation data.Experimental results based on a self-built UAV dataset show that this method can achieve high-precision detection without requiring annotation

providing an effective solution for smart power inspection in high-altitude and complex environmental conditions.