In the previous chapter, we learned how to use auxiliary information such as labels and attributes to improve the quality of images that are generated by GANs. The labels we used in the previous chapter were image-wise, which means that each image has only one or several labels. Labels can be assigned to specific pixels, which are known as pixel-wise labels. Pixel-wise labels are playing an increasingly important role in the realm of deep learning. For example, one of the most famous online image classification contests, the ImageNet Large Scale Visual Recognition Challenge (ILSVRC, http://www.image-net.org/challenges/LSVRC), is no longer being hosted since its last event in 2017, whereas object detection and segmentation challenges such as COCO (http://cocodataset.org) are receiving more attention.
An iconic application...
