Large-Scale and Fine-Grained Evaluation of Popular JPEG Forgery Localization Schemes

Abstract

Over the years, researchers have proposed various approaches to JPEG forgery detection and localization. In most cases, experimental evaluation was limited to JPEG quality levels that are multiples of 5 or 10. Each study used a different dataset, making it difficult to directly compare the reported results. The goal of this work is to perform a unified, large-scale and fine-grained evaluation of the most popular state-of-the-art detectors. The obtained results allow to compare the detectors with respect to various criteria, and shed more light on the compression configurations where reliable tampering localization can be expected.