A Self Supervised CNN for Image Watermark
Kothapally Deepthi · INTERANTIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING AND MANAGEMENT · 2025
Abstract - This project presents a self-supervised convolutional neural network (CNN) framework for robust image watermarking. Unlike traditional methods that rely on labeled data, our approach leverages self-supervised learning to embed and extract watermarks without explicit supervision. The network is trained to encode watermarks in imperceptible ways while maintaining high image quality and resistance to common distortions such as compression and noise. Experimental results demonstrate that our model achieves a strong balance between invisibility, robustness, and watermark retrieval accuracy. Key Words: Self-upervisedLearning,ConvolutionalNeuralNetwork(CNN),ImageWatermarking,DigitalWatermark,Robustness,Imperceptibility,Deep Learning,Data Hiding, Watermark Extraction,Image Processing