Graph Neural Network for Human Parsing

Human parsing, as a fine-grained semantic segmentation task, focuses on accurately identifying the various parts of the human body. It plays a crucial role in numerous downstream applications, such as crowd trajectory prediction, action recognition, virtual try-on, and virtual reality. With the advancements in deep Convolutional Neural Networks (CNNs), the performance of human parsing has witnessed significant improvements in recent years. It is worth noting that human bodies exhibit a natural topological physical structure, and different semantic body parts inherently exhibit correlations with one another. Hence, many methods apply graph neural network to this task.