Semantic segmentation is a type of computer vision technique that assigns a semantic label to each pixel in an image, such as "road," "car," "sky," or "building." By assigning each pixel a label, the computer can understand the contents of an image and distinguish between different objects in the image.