Image-FakeFinder : A Multi-Layered Forensic Framework for Deepfake and Manipulation Detection

The proliferation of deepfake technology and sophisticated image manipulation tools has significantly compromised the integrity of digital media. This paper introduces Image- FakeFinder, a comprehensive forensic framework designed to detect diverse forms of image forgery. Unlike conventional single-model detectors that often fail against novel generative techniques,Image-FakeFinder employs a multi-pillar approach integrating geometric, physical, biological, and sensor-level forensics. By combining Convolutional Neural Network (CNN) based heuristic analysis with Photo-Response Non- Uniformity (PRNU) fingerprinting and frequency-domain artifact detection, the system achieves high robustness across various datasets. Experimental results demonstrate that this hybrid methodology effectively identifies inconsistencies in light transport, facial geometry, and sensor noise patterns, providing a transparent and explainable verdict for digital forensics.