深度学习水下图像处理，将空气中的图片训练成水下图片

watergan 可以把空气中的正常图像通过gan网络生成水下图像

This paper reports on WaterGAN, a generative adversarial network (GAN) for generating realistic underwater imagesfromin-air image and depth pairings in an unsupervised pipeline used for color correction of monocular underwater images. Cameras onboard