دانلود کتاب روش ها و کاربردهای شبکه عصبی گراف در درک صحنه

This book mainly introduces the graph neural network and its applications in scene understanding. Graph neural network is developed for feature learning of graph data, which is the most prevalent form of data in the real world. This book aims to introduce the main existing techniques of graph neural network and explore some innovative applications of graph neural networks to scene understanding. It can offer a comprehensive guide to leveraging this powerful combination for advancing computer vision and beyond. This book is designed to serve as a comprehensive guide, explaining in detail how graph neural networks can be applied to speci_ic tasks in scene understanding and effectively solve the dif_iculties and challenges faced by the tasks. In general, this book is organized as follows: Chapter 1 introduces the history and frontiers of graph neural networks. Chapter 2 provides a detailed introduction to the task of scene understanding and highlights the three key tasks that this book focuses on. Chapter 3 explains the basic knowledge of graph neural network and the classical architectures of graph neural network in detail. Chapter 4 mainly introduces the graph convolutional network for scene parsing. Chapter 5 explores how graph neural networks can be applied to both strongly supervised and weakly supervised human parsing. Chapter 6 introduces the dynamic graph construction for human parsing and the dynamic graph pooling for human parsing. Chapter 7 introduces the graph neural networks for video object segmentation. Chapter 8 summarizes the book and proposes some future research directions for scene understanding tasks.