Deep Learning for Visible Watermark Removal: A Survey
Peixian Su, Yong Zhang · Computational Intelligence · 2025
ABSTRACT With the advancement of deep learning technology, deep learning methods are increasingly applied to image restoration, especially in the field of visible watermark removal from images. These methods play an important role and have achieved remarkable success. However, there is a scarcity of literature summarizing the application of different deep learning methods in the field of image watermark removal. In this paper, we present a comparative study of image watermark removal methods from different perspectives. First, we take a look at the development of image restoration techniques. Second, we present the popular architectures of deep learning networks for image applications. Then, we analyze deep learning‐based watermark removal methods from both supervised and unsupervised perspectives and provide insights into the motivation and principle of various deep learning methods, which will be analyzed by integrating different network architectures and methodological frameworks. Thirdly, we compare the performance of these popular watermark removal methods on public watermarked datasets in terms of quantitative and qualitative analysis. Finally, we highlight the challenges and potential research directions of current watermarking methods. We review and summarize deep learning‐based methods for visible watermark removal, aiming to help evaluate existing removal techniques and advance the field of image watermarking.