WSDS-GAN: A weak-strong dual supervised learning method for underwater image enhancement

•The difficult of acquiring large high-quality underwater image hinders the development of the deep learning-based underwater image enhancement method.•WSDS-GAN conducts the style transfer from degradation style to clear style for underwater image enhancement based on CycleGAN.•The two-stage learning strategy solves the problem of big data deficiency and improves the enhanced image quality efficiently in a single framework.•The performance of WSDS-GAN is competitive with other state-of-the-art underwater image enhancement methods in public datasets.•WSDS-GAN performs well both in real underwater images and application assessment.