Semantic segmentation is a classification task in computer vision that assigns a class to each pixel of an image, effectively segmenting the image into regions of interest. For example, the Cambridge-driving Labeled Video Database (CamVid) is a dataset that includes a collection of videos recorded from the perspective of a driving car, with over 700 frames that have been labeled to assign one of 32 semantic classes (e.g. pedestrian, bicyclist, road, etc.) to each pixel. This dataset can be used to train a model to segment new scenes into those same classes.