The encoder downsamples the input image to grasp its intricate details, while the decoder upsamples to reconstruct it, maintaining consistency with UNet’s proven advantages.