Fig. 4
From: CPDM: Content-preserving diffusion model for underwater image enhancement
Visual comparison of enhanced underwater images by various methods on the Test_L400 (LSUI) dataset. The leftmost Input represents the underwater image that needs to be enhanced, while the rightmost GT represents the clean reference image. The methods we compared include WaterNet43, FUnIE57, Ucolor25, U-shape Transformer28, TBDM32, WWPF34, and DiffWater29.
