FPGA-based real-time image tampering detection system for edge computing

With the proliferation of sophisticated image manipulation techniques, robust and efficient tampering detection has become critically important for multimedia forensics and security applications. Although existing GPU or CPU-based deep learning detection methods are effective, they often rely on cloud services for data transmission, which introduces latency, bandwidth overhead, and privacy vulnerabilities. Edge computing provides a promising solution to this challenge. This paper presents a highly energy-efficient real-time image tampering detection system for edge computing environments. The system achieves significant energy efficiency improvements while maintaining detection accuracy by employing a novel hardware-software co-optimization approach. Experimental results demonstrate that the proposed system outperforms comparable GPU and CPU implementations by 1.99× and 122.07× in energy efficiency, respectively.