Abstract: The sustained advancement of image processing technology makes digital image editing more and more easy. With popular image processing software, people can easily manipulate image contents. The manipulated images are becoming so realistic that they are difficult to be identified with the naked eyes, which have posed serious threats to personal privacy, social order, and even national security. Therefore, it is of great importance to detect and localize the tampered regions in digital images, which has attracted much attention in the field of multimedia information security. In recent years, deep learning technology has been widely adopted in image tampering localization and has significantly outperformed traditional forensic methods. This paper reviews the image tampering localization methods based on deep learning. It introduces the commonly used datasets and evaluation criteria for image tampering localization. Based on the applications of different network architectures, the technical features and localization performance of existing methods are presented. In addition, the challenges of image tampering localization and future research directions are discussed.