Abstract:Steganography and digital watermarking are the tasks of hiding recoverable data in image pixels. Deep neural network (DNN) based image steganography and watermarking techniques are quickly replacing traditional hand-engineered pipelines. DNN based watermarking techniques have drastically improved the message capacity, imperceptibility and robustness of the embedded watermarks. However, this improvement comes at the cost of increased computational overhead of the watermark encoder neural network. In this work, we design the first accelerator platform FastStamp to perform DNN based steganography and digital watermarking of images on hardware. We first propose a parameter efficient DNN model for embedding recoverable bit-strings in image pixels. Our proposed model can match the success metrics of prior state-of-the-art DNN based watermarking methods while being significantly faster and lighter in terms of memory footprint. We then design an FPGA based accelerator framework to further improve the model throughput and power consumption by leveraging data parallelism and customized computation paths. FastStamp allows embedding hardware signatures into images to establish media authenticity and ownership of digital media. Our best design achieves 68 times faster inference as compared to GPU implementations of prior DNN based watermark encoder while consuming less power.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the large computational overhead, high memory requirements, and difficulty in deploying on resource - constrained hardware when embedding and extracting watermarks in existing deep - learning - based image steganography and digital watermarking techniques. Specifically: 1. **Improve efficiency**: Although traditional hand - engineered pipeline methods can achieve image steganography and digital watermarking, these methods usually result in high - visibility of the embedded watermark, limited message capacity, and poor robustness to digital compression techniques (such as JPEG transformation). While methods based on deep neural networks (DNN) have significantly improved message capacity, invisibility, and robustness, these improvements come at the cost of increased computational overhead and memory requirements. Therefore, the paper proposes a parameter - efficient DNN model, aiming to reduce computational overhead and memory usage while maintaining high performance. 2. **Hardware acceleration**: Existing DNN watermark embedding methods are mainly implemented at the software level, resulting in a large delay between image capture and transmission. To achieve real - time image and video watermark embedding, the paper designs an FPGA acceleration framework, which further improves the model's throughput and reduces power consumption by leveraging data parallelism and customized computational paths. 3. **Support multiple watermarking schemes**: The FastStamp framework proposed in the paper not only supports robust watermarks (for media ownership and copyright protection) but also supports semi - fragile watermarks (for checking media integrity and authenticity). Semi - fragile watermarks need to maintain robustness in benign image - processing operations (such as compression, scaling, and color adjustment), while being sensitive to malicious operations (such as image tampering and face - swapping). 4. **Media authentication**: With the development of deep - learning - generation technologies, it has become increasingly easy to generate high - quality fake images and videos, which may lead to the spread of false information and a trust crisis. The method proposed in the paper can embed a unique hardware signature at the time of image capture, thereby providing a means of verifying media authenticity and preventing media forgery and tampering. In summary, the main objective of this paper is to design a lightweight and efficient neural image watermarking framework that can achieve real - time image watermark embedding on hardware platforms and support multiple watermarking schemes to meet the requirements of different application scenarios.

FastStamp: Accelerating Neural Steganography and Digital Watermarking of Images on FPGAs

FPGA based fast implementation of RS steganalytic algorithm

High-performance Reconfigurable DNN Accelerator on a Bandwidth-limited Embedded System

DeepHardMark: Towards Watermarking Neural Network Hardware

FPGA-based Hardware Acceleration for Image Copyright Protection Syetem Based on Blockchain

Design and Implementation of Digital Image Watermark Based on FPGA

FastWave: Accelerating Autoregressive Convolutional Neural Networks on FPGA

A Hidden DCT-Based Invisible Watermarking Method for Low-Cost Hardware Implementations

WPU: A FPGA-based Scalable, Efficient and Software/Hardware Co-design Deep Neural Network Inference Acceleration Processor

A High-Performance Accelerator for Real-Time Super-Resolution on Edge FPGAs

Feasibility of an Encryption and Decryption System for Messages and Images Using a Field Programmable Gate Array (FPGA) As the Portable Encryption Key Platform

FaceSigns: Semi-Fragile Watermarks for Media Authentication

A Design Methodology for Efficient Implementation of Deconvolutional Neural Networks on an FPGA

Deepfake attack prevention using steganography GANs

A FPGA-based Accelerator of Convolutional Neural Network for Face Feature Extraction

Toward Full-Stack Acceleration of Deep Convolutional Neural Networks on FPGAs

Optimizing FPGA-based Accelerator Design for Deep Convolutional Neural Networks

HARDWARE ACCELERATOR: IMPLEMENTATION OF CNN ON FPGA FOR DIGIT RECOGNITION

FPGA-based Accelerator for Convolutional Neural Network

FPGA-Based Feature Extraction and Tracking Accelerator for Real-Time Visual SLAM

Embedded Streaming Deep Neural Networks Accelerator With Applications