Pixel-level annotation is an annotation method that takes a single pixel in an image as the minimum annotation unit, and annotates each pixel with category, attribute or semantic information (such as annotations in semantic segmentation and panoptic segmentation), realizing the finest-grained information marking of images, used for high-precision image segmentation and pixel-level recognition model training.
