This paper reviews current and future trends in deep learning-based watermarking techniques for images and videos. It discusses the challenges and developments in both traditional and deep learning-based watermarking methods. Traditional watermarking techniques are based on embedding signatures into specific domains, such as spatial or frequency domains, and are effective in terms of invisibility and robustness. However, deep learning-based methods have shown greater efficiency in embedding and extracting watermarks, especially in image watermarking, where the network architecture significantly affects the invisibility and robustness of the watermark. For video watermarking, deep learning-based techniques are still in their early stages, with limited research focusing on this area. The paper highlights the importance of robustness, invisibility, and capacity in watermarking systems and discusses the challenges of deep learning-based video watermarking, including the need for robustness against various attacks and the difficulty of embedding watermarks in video content. The paper also presents a classification of deep learning-based watermarking techniques based on network architecture and embedding domain, and compares the most popular techniques in terms of their performance. It concludes by discussing future directions in deep learning-based video watermarking, emphasizing the need for further research in this area.This paper reviews current and future trends in deep learning-based watermarking techniques for images and videos. It discusses the challenges and developments in both traditional and deep learning-based watermarking methods. Traditional watermarking techniques are based on embedding signatures into specific domains, such as spatial or frequency domains, and are effective in terms of invisibility and robustness. However, deep learning-based methods have shown greater efficiency in embedding and extracting watermarks, especially in image watermarking, where the network architecture significantly affects the invisibility and robustness of the watermark. For video watermarking, deep learning-based techniques are still in their early stages, with limited research focusing on this area. The paper highlights the importance of robustness, invisibility, and capacity in watermarking systems and discusses the challenges of deep learning-based video watermarking, including the need for robustness against various attacks and the difficulty of embedding watermarks in video content. The paper also presents a classification of deep learning-based watermarking techniques based on network architecture and embedding domain, and compares the most popular techniques in terms of their performance. It concludes by discussing future directions in deep learning-based video watermarking, emphasizing the need for further research in this area.