Abstract: Hydrophobicity is an important indicator for judging the performance of insulators, and identifying the hydrophobicity of insulators is time-consuming, labor-intensive, and has unstable performance. This article proposes a hydrophobicity detection method based on the FasterNet model and ensemble learning soft voting strategy. Firstly, three variants are constructed based on the FasterNet classification network to form three base detectors. Then, the hydrophobicity prediction results of the base detectors are subjected to ensemble learning soft voting to obtain the optimal detection result. The experiment shows that the method proposed in this paper effectively improves the accuracy and stability of insulator hydrophobicity detection.