Image Forgery Detection
Overview
Image forgery detection asks whether an image is authentic, what was changed, and where the changes occurred, across settings such as face deepfakes, edited photographs, and forged documents. Methods look for spatial, frequency, noise, and compression artifacts, while newer systems adapt pretrained vision models and multimodal language models to combine forensic cues with semantic reasoning. Benchmarks test image-level classification and pixel-level localization, increasingly across unseen generators, datasets, and realistic degradation. A central challenge is that detectors can rely on dataset shortcuts rather than forgery evidence, so cross-domain robustness, calibrated decisions, and grounded explanations remain open concerns.
Latest papers
127 papers12 papers in the last four weeks, down 25% on the four weeks before. 0.1% of all new papers.