A Survey of Watermarking Technique using Deep Neural Network Architecture
Megha Gupta, R. Rama Kishore · 2021 International Conference on Computing, Communication, and Intelligent Systems (ICCCIS) · 2021
A computerized watermark is a sort of marker secretly introduced in a noise accepting signals, let us say, video, audio or image. It is normally accustomed to recognize proprietorship for the copyright of such signals. "Watermarking" is the way toward covering computerized bits in a conveyor signal. The shrouded data need not have to contain a connection to the transporter signal. Computerized watermarks might be applied to verify the genuineness or uprightness of the conveyor signal or to demonstrate the uniqueness of its proprietors. This paper explores a new family of transformation dependent on Deep Learning systems. We survey and present a relative investigation of the different profound learning procedures that target installing watermark that can be worked upon to accomplish substantial security level for the data being transmitted, and robustness over a few attacks particularly when transmitted over the loud medium. Our paper centers on how deep learning-based watermarking approaches show their prevalence because of the level of imperceptibility and robustness presented by adjustments in recurrence coefficients. We likewise feature the parts of the current methodologies that can be worked upon to accomplish images that are exceptionally indistinct from one another in the future.