U Net Architecture

U-Net is a convolutional neural network architecture primarily used for image segmentation tasks, aiming to accurately delineate regions of interest within an image. Current research focuses on adapting and improving U-Net's performance through modifications like incorporating attention mechanisms, adding skip connections with embedded physical models, and integrating it with other architectures such as EfficientNet or transformers. These advancements enhance accuracy and efficiency across diverse applications, including medical image analysis (e.g., tumor segmentation, retinal vessel extraction), remote sensing (e.g., building extraction), and signal processing (e.g., phonocardiogram denoising), demonstrating its broad utility and impact.