Title: Autoencoder-CNN-based embedding and extraction model for image watermarking
Published in: Journal of Electronic Imaging, SPIE | Volume 32, Issue 2

→ Objective: This research is part of my foundational work in applied AI—focusing on model architecture, learning dynamics, and image-based intelligence systems before the rise and surge of usage of generative AI tools.

→ Thought and Research Process:

• Intention – This work emerged from a desire to create a robust, intelligent system capable of hiding and retrieving data within digital images without loss, applying core principles of deep learning beyond surface-level tool use.

• Content – This deep learning model hides one image inside another—like embedding an image watermark (from the MNIST dataset) invisibly into a completely different photo. The model can later extract the hidden image [watermark] accurately, even if the outer image is compressed or slightly distorted. 

• Impact – The model offered a more intelligent and secure approach to data hiding, relevant to domains requiring privacy, integrity, and reversibility. The robustness has been tested across different attacks. This technique offers a powerful way to protect digital image ownership using AI.

⚙️ Technical Highlights:
1. Neural Networks Used: Convolutional Neural Networks (CNNs), Autoencoders
2. Architecture: Two-part system (embedding + extraction models)
3. Tech Stack: Python, TensorFlow
4. Key Techniques: Feature map fusion, transposed convolutions, ReLU activation
5. Performance: High invisibility and robustness; outperformed other methods under multiple distortion attacks