A comprehensive review of video watermarking technique in deep learning environments
Souha Mansour, Saoussen Ben Jabra, Ezzeddine Zagrouba · 2023
In recent years, the advent of the Internet and the rapid growth of digital media applications have made video a primary medium for information transmission. However, this progress has also brought about new challenges, including concerns related to unauthorized copying, digital plagiarism, and the distribution and utilization of copyrighted digital content. In order to address these issues, watermarking has emerged as a solution. It involves embedding a watermark into a digital cover and subsequently extracting it to resolve ownership disputes and copyright infringements related to media content. Numerous conventional video watermarking techniques have been introduced, demonstrating their effectiveness in achieving both invisibility and robustness against various types of attacks. Recently, the application of deep learning principles in embedding signatures into video content has gained significant attention. This approach offers considerable advantages in the field of watermarking due to its accuracy, superior outcomes, and exceptional learning capabilities. This paper provides an overview of recent advancements in deep learning-based video watermarking. It categorizes the proposed approaches according to the employed network architecture, offering a comprehensive summary of the field’s latest developments. The study concludes by examining potential research avenues in the domain of deep learning-based video watermarking.