Visualization of the segmentations results from different fully supervised methods. The UNet, 3D UNet, UNet++, and 3D ResNet models are commonly employed for medical image segmentation tasks, while RetiFluidNet is specifically designed for 2D segmentation of retinal fluid. These models typically undergo supervised training utilizing all annotated datasets. In our approach, we adopt VNet as the backbone network and employ only 25% of the annotated data for training purposes.