What is U-Net?

U-Net is a convolutional neural network architecture primarily used for image segmentation. It consists of a contracting path (encoder) to capture context and a symmetric expanding path (decoder) to enable precise localization. Skip connections between the encoder and decoder paths help to recover spatial information lost during downsampling. It's widely used in medical image analysis, satellite imagery analysis, and other applications requiring pixel-level classification.