I have included a new model called U-NeXt, which is a variation of U-Net that uses ConvNext blocks at each level. I have made this changes:
blocks.py: I have added the ConvNextBlock_v1 and the UpConvNextBlock_V1 functions. This second function does the upsampling and uses the first function to integrate ConvNext blocks in each upsampling layer.
unext_v1.py: U-NeXt architecture implementation.
config.py: Three new variables were included: MODEL.CONVNEXT_LAYERS (a list of the number of ConvNext blocks in each level), MODEL.CONVNEXT_SD_PROB (maximum stochastic probability) and MODEL.CONVNEXT_LAYER_SCALE (layer scale parameter).
check_configuration.py: Included 'unext_v1' in relevant sections.
init.py: The U-NeXt architecture was included with corresponded arguments.
I have included a new model called U-NeXt, which is a variation of U-Net that uses ConvNext blocks at each level. I have made this changes: