Successful forensic detectors can produce excellent results in supervised learning benchmarks but struggle to transfer to real-world applications. We believe this limitation is largely due to inadequate training data quality. While most research focuses on developing new algorithms, less attention is given to training data selection, despite evidence that performance can be strongly impacted by spurious correlations such as content, format, or resolution. A well-designed forensic detector should detect generator specific artifacts rather than reflect data biases. To this end, we propose B-Free, a bias-free training paradigm, where fake images are generated from real ones using the conditioning procedure of stable diffusion models. This ensures semantic alignment between real and fake images, allowing any differences to stem solely from the subtle artifacts introduced by AI generation. Through content-based augmentation, we show significant improvements in both generalization and robustness over state-of-the-art detectors and more calibrated results across 27 different generative models, including recent releases, like FLUX and Stable Diffusion 3.5. Our findings emphasize the importance of a careful dataset design, highlighting the need for further research on this topic. Code and data are publicly available at https://grip-unina.github.io/B-Free/.

The rapid advancement of generative AI has revolutionized image creation, enabling high-quality synthesis from text prompts while raising critical challenges for media authenticity. We present AI-GenBench, a novel benchmark designed to address the urgent need for robust detection of AI-generated images in real-world scenarios. Unlike existing solutions that evaluate models on static datasets, AI-GenBench introduces a temporal evaluation framework where detection methods are incrementally trained on synthetic images, historically ordered by their generative models, to test their ability to generalize to new generative models, such as the transition from GANs to diffusion models. Our benchmark focuses on high-quality, diverse visual content and overcomes key limitations of current approaches, including arbitrary dataset splits, unfair comparisons, and excessive computational demands. AI-GenBench provides a comprehensive dataset, a standardized evaluation protocol, and accessible tools for both researchers and non-experts (e.g., journalists, fact-checkers), ensuring reproducibility while maintaining practical training requirements. By establishing clear evaluation rules and controlled augmentation strategies, AI-GenBench enables meaningful comparison of detection methods and scalable solutions. Code and data are publicly available to ensure reproducibility and to support the development of robust forensic detectors to keep pace with the rise of new synthetic generators1

Synthetic video generation is progressing very rapidly. The latest models can produce very realistic high-resolution videos that are virtually indistinguishable from real ones. Although several video forensic detectors have been recently proposed, they often exhibit poor generalization, which limits their applicability in a real-world scenario. Our key insight to overcome this issue is to guide the detector towards _seeing_ _what_ _really_ _matters_. In fact, a well-designed forensic classifier should focus on identifying intrinsic low-level artifacts introduced by a generative architecture rather than relying on high-level semantic flaws that characterize a specific model. In this work, first, we study different generative architectures, searching and identifying discriminative features that are unbiased, robust to impairments, and shared across models. Then, we introduce a novel forensic-oriented data augmentation strategy based on the wavelet decomposition and replace specific frequency-related bands to drive the model to exploit more relevant forensic cues. Our novel training paradigm improves the generalizability of AI-generated video detectors, without the need for complex algorithms and large datasets that include multiple synthetic generators. To evaluate our approach, we train the detector using data from a single generative model and test it against videos produced by a wide range of other models. Despite its simplicity, our method achieves a significant accuracy improvement over state-of-the-art detectors and obtains excellent results even on very recent generative models, such as NOVA and FLUX.

The ability to detect manipulated visual content is becoming increasingly important in many application fields, given the rapid advances in image synthesis methods. Of particular concern is the possibility of modifying the content of medical images, altering the resulting diagnoses. Despite its relevance, this issue has received limited attention from the research community. One reason is the lack of large and curated datasets to use for development and benchmarking purposes. Here, we investigate this issue and propose M3Dsynth, a large dataset of manipulated Computed Tomography (CT) lung images. We create manipulated images by injecting or removing lung cancer nodules in real CT scans, using three different methods based on Generative Adversarial Networks (GAN) or Diffusion Models (DM), for a total of 8,577 manipulated samples. Experiments show that these images easily fool automated diagnostic tools. We also tested several state-of-the-art forensic detectors and demonstrated that, once trained on the proposed dataset, they are able to accurately detect and localize manipulated synthetic content, even when training and test sets are not aligned, showing good generalization ability. Dataset and code are publicly available at https://grip-unina.github.io/M3Dsynth/.

Generalization is a main issue for current audio deepfake detectors, which struggle to provide reliable results on out-of-distribution data. Given the speed at which more and more accurate synthesis methods are developed, it is important to design techniques that work well also on data they were not trained for. In this paper we study the potential of large-scale pre-trained models for audio deepfake detection. To this end, the detection problem is reformulated as a speaker verification framework and fake audios are exposed by the mismatch between the voice sample under test and the voice of the claimed identity. With this paradigm, no fake speech sample is necessary in training, cutting off any link with the generation method at the root. Features are extracted by general-purpose large pre-trained models, with no need for training or fine-tuning on specific datasets. At detection time only a limited set of voice fragments of the identity under test is required. Experiments on several datasets show that detectors based on pre-trained models achieve excellent performance and show strong generalization ability, rivaling supervised methods on in-distribution data and largely overcoming them on out-of-distribution data.

High-quality artificially generated images are widely available now and increasingly realistic, posing challenges for image forensics in distinguishing them from real ones. Unfortunately, building a single detector that generalizes well to unseen generators is very difficult, creating the need for diverse cues. In this paper, we show that natural and synthetic images differ in their color statistics, possibly due to the widely used perceptual loss, which is more sensitive to brightness than to chroma differences. Consequently, color statistics offer valuable cues for forensic analysis and the development of robust detectors. Our experiments using simple hand-crafted color functions with a random forest achieve 91% accuracy averaged over all tested Diffusion Models, even with limited training samples.

In this paper we present TruFor, a forensic framework that can be applied to a large variety of image manipulation methods, from classic cheapfakes to more recent manipulations based on deep learning. We rely on the extraction of both high-level and low-level traces through a transformer-based fusion architecture that combines the RGB image and a learned noise-sensitive fingerprint. The latter learns to embed the artifacts related to the camera internal and external processing by training only on real data in a self-supervised manner. Forgeries are detected as deviations from the expected regular pattern that characterizes each pristine image. Looking for anomalies makes the approach able to robustly detect a variety of local manipulations, ensuring generalization. In addition to a pixel-level localization map and a whole-image integrity score, our approach outputs a reliability map that highlights areas where localization predictions may be error-prone. This is particularly important in forensic applications in order to reduce false alarms and allow for a large scale analysis. Extensive experiments on several datasets show that our method is able to reliably detect and localize both cheapfakes and deepfakes manipulations outperforming state-of-the-art works. Code is publicly available at https://grip-unina.github.io/TruFor/

Over the past decade, there has been tremendous progress in creating synthetic media, mainly thanks to the development of powerful methods based on generative adversarial networks (GAN). Very recently, methods based on diffusion models (DM) have been gaining the spotlight. In addition to providing an impressive level of photorealism, they enable the creation of text-based visual content, opening up new and exciting opportunities in many different application fields, from arts to video games. On the other hand, this property is an additional asset in the hands of malicious users, who can generate and distribute fake media perfectly adapted to their attacks, posing new challenges to the media forensic community. With this work, we seek to understand how difficult it is to distinguish synthetic images generated by diffusion models from pristine ones and whether current state-of-the-art detectors are suitable for the task. To this end, first we expose the forensics traces left by diffusion models, then study how current detectors, developed for GAN-generated images, perform on these new synthetic images, especially in challenging social-network scenarios involving image compression and resizing. Datasets and code are available at https:github.com/grip-unina/DMimageDetection.

A major challenge in DeepFake forgery detection is that state-of-the-art algorithms are mostly trained to detect a specific fake method. As a result, these approaches show poor generalization across different types of facial manipulations, e.g., from face swapping to facial reenactment. To this end, we introduce ID-Reveal, a new approach that learns temporal facial features, specific of how a person moves while talking, by means of metric learning coupled with an adversarial training strategy. The advantage is that we do not need any training data of fakes, but only train on real videos. Moreover, we utilize high-level semantic features, which enables robustness to widespread and disruptive forms of post-processing. We perform a thorough experimental analysis on several publicly available benchmarks. Compared to state of the art, our method improves generalization and is more robust to low-quality videos, that are usually spread over social networks. In particular, we obtain an average improvement of more than 15% in terms of accuracy for facial reenactment on high compressed videos.

The advent of deep learning has brought a significant improvement in the quality of generated media. However, with the increased level of photorealism, synthetic media are becoming hardly distinguishable from real ones, raising serious concerns about the spread of fake or manipulated information over the Internet. In this context, it is important to develop automated tools to reliably and timely detect synthetic media. In this work, we analyze the state-of-the-art methods for the detection of synthetic images, highlighting the key ingredients of the most successful approaches, and comparing their performance over existing generative architectures. We will devote special attention to realistic and challenging scenarios, like media uploaded on social networks or generated by new and unseen architectures, analyzing the impact of suitable augmentation and training strategies on the detectors’ generalization ability.

The ever higher quality and wide diffusion of fake images have spawn a quest for reliable forensic tools. Many GAN image detectors have been proposed, recently. In real world scenarios, however, most of them show limited robustness and generalization ability. Moreover, they often rely on side information not available at test time, that is, they are not universal. We investigate these problems and propose a new GAN image detector based on a limited sub-sampling architecture and a suitable contrastive learning paradigm. Experiments carried out in challenging conditions prove the proposed method to be a first step towards universal GAN image detection, ensuring also good robustness to common image impairments, and good generalization to unseen architectures.

Abstract PRNU-based image processing is a key asset in digital multimedia forensics. It allows for reliable device identification and effective detection and localization of image forgeries, in very general conditions. However, performance impairs significantly in challenging conditions involving low quality and quantity of data. These include working on compressed and cropped images or estimating the camera PRNU pattern based on only a few images. To boost the performance of PRNU-based analyses in such conditions, we propose to leverage the image noiseprint, a recently proposed camera-model fingerprint that has proved effective for several forensic tasks. Numerical experiments on datasets widely used for source identification prove that the proposed method ensures a significant performance improvement in a wide range of challenging situations.

Source identification is an important topic in image forensics, since it allows to trace back the origin of an image. This represents a precious information to claim intellectual property but also to reveal the authors of illicit materials. In this letter we address the problem of device identification based on sensor noise and propose a fast and accurate solution using convolutional neural networks (CNNs). Specifically, we propose a 2-channel-based CNN that learns a way of comparing camera fingerprint and image noise at patch level. The proposed solution turns out to be much faster than the conventional approach and to ensure an increased accuracy. This makes the approach particularly suitable in scenarios where large databases of images are analyzed, like over social networks. In this vein, since images uploaded on social media usually undergo at least two compression stages, we include investigations on double JPEG compressed images, always reporting higher accuracy than standard approaches.

Forensic analyses of digital images rely heavily on the traces of in-camera and out-camera processes left on the acquired images. Such traces represent a sort of camera fingerprint. If one is able to recover them, by suppressing the high-level scene content and other disturbances, a number of forensic tasks can be easily accomplished. A notable example is the PRNU pattern, which can be regarded as a device fingerprint, and has received great attention in multimedia forensics. In this paper, we propose a method to extract a camera model fingerprint, called noiseprint, where the scene content is largely suppressed and model-related artifacts are enhanced. This is obtained by means of a Siamese network, which is trained with pairs of image patches coming from the same (label +1) or different (label -1) cameras. Although the noiseprints can be used for a large variety of forensic tasks, in this paper we focus on image forgery localization. Experiments on several datasets widespread in the forensic community show noiseprint-based methods to provide state-of-the-art performance.

The rapid progress in synthetic image generation and manipulation has now come to a point where it raises significant concerns for the implications towards society. At best, this leads to a loss of trust in digital content, but could potentially cause further harm by spreading false information or fake news. This paper examines the realism of state-of-the-art image manipulations, and how difficult it is to detect them, either automatically or by humans. To standardize the evaluation of detection methods, we propose an automated benchmark for facial manipulation detection. In particular, the benchmark is based on Deep-Fakes, Face2Face, FaceSwap and NeuralTextures as prominent representatives for facial manipulations at random compression level and size. The benchmark is publicly available and contains a hidden test set as well as a database of over 1.8 million manipulated images. This dataset is over an order of magnitude larger than comparable, publicly available, forgery datasets. Based on this data, we performed a thorough analysis of data-driven forgery detectors. We show that the use of additional domain-specific knowledge improves forgery detection to unprecedented accuracy, even in the presence of strong compression, and clearly outperforms human observers.

Nonlocal methods are state-of-the-art in SAR despeckling, thanks to their ability to exploit image self-similarity. Given sufficient training data, however, methods based on deep learning have proven highly competitive. Therefore, to take the best of both approaches, we investigate the use of deep learning to improve nonlocal despeckling. We use plain non-iterative nonlocal means despeckling, with weights provided, for each estimation window, by a suitably trained deep CNN. Experiments on synthetic and real SAR data prove this approach to outperform conventional nonlocal methods.

We propose a new algorithm for the reliable detection and localization of video copy–move forgeries. Discovering well-crafted video copy–moves may be very difficult, especially when some uniform background is copied to occlude foreground objects. To reliably detect both additive and occlusive copy–moves, we use a dense-field approach, with invariant features that guarantee robustness to several postprocessing operations. To limit complexity, a suitable video-oriented version of PatchMatch is used, with a multiresolution search strategy, and a focus on volumes of interest. Performance assessment relies on a new dataset, designedad hoc, with realistic copy–moves and a wide variety of challenging situations. Experimental results show the proposed method to detect and localize video copy–moves with good accuracy even in adverse conditions.

We propose a new method for synthetic aperture radar (SAR) image despeckling, which leverages information drawn from coregistered optical imagery. Filtering is performed by patchwise nonlocal means, working exclusively on SAR data. However, the filtering weights are computed by taking into account also the optical guide, which is much cleaner than the SAR image, and hence more discriminative. To avoid injecting optical-domain information into the filtered image, an SAR-domain statistical test is preliminarily performed to reject right away any risky predictor. Experiments on two SAR-optical data sets prove the proposed method to suppress very effectively the speckle, preserving structural details, and without introducing significant filtering artifacts. Overall, the proposed method compares favorably with all the state-of-the-art despeckling filters, and also with our own previous optical-guided filter.

Nonlocal filters show outstanding performance in the field of interferometric phase restoration by providing strong filtering power together with high spatial features preservation. In this work we focus on the generation of Digital Elevation Models (DEM) from a pair of interferometric SAR images. In the specific, we aim at comparing the performance of state-of-the-art InSAR filtering approaches on the basis of their noise suppression and detail preservation capabilities. We exploit a dataset of TanDEM-X SAR data relative to the volcanic area of the Kamchatka region (Russia).

We recently proposed a convolutional neural network (CNN) for remote sensing image pansharpening obtaining a significant performance gain over the state of the art. In this paper, we explore a number of architectural and training variations to this baseline, achieving further performance gains with a lightweight network that trains very fast. Leveraging on this latter property, we propose a target-adaptive usage modality that ensures a very good performance also in the presence of a mismatch with respect to the training set and even across different sensors. The proposed method, published online as an off-the-shelf software tool, allows users to perform fast and high-quality CNN-based pansharpening of their own target images on general-purpose hardware.

The block-matching 3-D (BM3D) algorithm, based on the nonlocal approach, is one of the most effective methods to date for additive white Gaussian noise image denoising. Likewise, its extension to synthetic aperture radar (SAR) amplitude images, SAR-BM3D, is a state-of-the-art SAR despeckling algorithm. In this paper, we further extend BM3D to address the restoration of SAR interferometric phase images. While keeping the general structure of BM3D, its processing steps are modified to take into account the peculiarities of the SAR interferometry signal. Experiments on simulated and real-world Tandem-X SAR interferometric pairs prove the effectiveness of the proposed method.

Video content can be acquired with off-the-shelf hardware, and is thus increasingly used to record events. With the growing role of video data for communicating to a large audience, we need tools to ensure the authenticity of video content. However, until now, only few methods exist to forensically analyze videos. In this work, we propose a method for statistically comparing two video sequences. Per sequence, intra- and inter-frame residuals are computed. Optical flow is used to compensate for motion artifacts on inter-frame residuals. We use one sequence to build a statistical model, and compare it to the second sequence. From a forensic perspective, the proposed method enables two applications. First, manipulations can be accurately localized if both sequences are subsequences of the same video. Second, source cameras can be distinguished if both sequences stem from different videos. The proposed method is evaluated on collected smartphone data and green-screen splices. Further, it is quantitatively compared to both a recent PRNU-based approach and a technique based on autoencoders.

In this paper we investigate the use of discriminative model learning through Convolutional Neural Networks (CNNs) for SAR image despeckling. The network uses a residual learning strategy, hence it does not recover the filtered image, but the speckle component, which is then subtracted from the noisy one. Training is carried out by considering a large multitemporal SAR image and its multilook version, in order to approximate a clean image. Experimental results, both on synthetic and real SAR data, show the method to achieve better performance with respect to state-of-the-art techniques.

Ship detection is a fundamental task for SAR-based maritime surveillance. Besides providing high reliability, a good detector is required to be computationally light, in order to analyze huge areas in a reasonable time. We propose a fully convolutional neural network for ship detection in SAR images. Thanks to a relatively simple architecture, complexity remains low enough to allow for a single-stage approach, thus avoiding the possible errors of CFAR pre-screening. Experiments on a Sentinel-1 dataset prove the proposed CNN to be much more reliable than CFAR detection.

This work deals with the fusion of SAR and optical data for land cover monitoring. We first propose to use co-registered optical data as a guide for nonlocal SAR image despeckling. Then, we fuse filtered Sentinel-1 SAR data with optical Sentinel-2 data for land-use classification. Experiments show that using optical-driven despeckled SAR data largely improves classification accuracy w.r.t using the original data and even multitemporal filtered data.

Local descriptors based on the image noise residual have proven extremely effective for a number of forensic applications, like forgery detection and localization. Nonetheless, motivated by promising results in computer vision, the focus of the research community is now shifting on deep learning. In this paper we show that a class of residual-based descriptors can be actually regarded as a simple constrained convolutional neural network (CNN). Then, by relaxing the constraints, and fine-tuning the net on a relatively small training set, we obtain a significant performance improvement with respect to the conventional detector.

We propose a new algorithm for fast approximate nearest neighbor search based on the properties of ordered vectors. Data vectors are classified based on the index and sign of their largest components, thereby partitioning the space in a number of cones centered in the origin. The query is itself classified, and the search starts from the selected cone and proceeds to neighboring ones. Overall, the proposed algorithm corresponds to locality sensitive hashing in the space of directions, with hashing based on the order of components. Thanks to the statistical features emerging through ordering, it deals very well with the challenging case of unstructured data, and is a valuable building block for more complex techniques dealing with structured data. Experiments on both simulated and real-world data prove the proposed algorithm to provide a state-of-the-art performance.

We propose a new algorithm for the accurate detection and localization of copy-move forgeries, based on rotation-invariant features computed densely on the image. Dense-field techniques proposed in the literature guarantee a superior performance with respect to their keypoint-based counterparts, at the price of a much higher processing time, mostly due to the feature matching phase. To overcome this limitation, we resort here to a fast approximate nearest-neighbor search algorithm, PatchMatch, especially suited for the computation of dense fields over images. We adapt the matching algorithm to deal efficiently with invariant features, so as to achieve higher robustness with respect to rotations and scale changes. Moreover, leveraging on the smoothness of the output field, we implement a simplified and reliable postprocessing procedure. The experimental analysis, conducted on databases available online, proves the proposed technique to be at least as accurate, generally more robust, and typically much faster than the state-of-the-art dense-field references.

PRNU-based techniques guarantee a good forgery detection performance irrespective of the specific type of forgery. The presence or absence of the camera PRNU pattern is detected by a correlation test. Given the very low power of the PRNU signal, however, the correlation must be averaged over a pretty large window, reducing the algorithm's ability to reveal small forgeries. To improve resolution, we estimate correlation with a spatially adaptive filtering technique, with weights computed over a suitable pilot image. Implementation efficiency is achieved by resorting to the recently proposed guided filters. Experiments prove that the proposed filtering strategy allows for a much better detection performance in the case of small forgeries.

We propose a new image forgery detection technique which fuses the outputs of two very diverse tools, based on machine learning and block-matching, respectively. The machine-learning tool builds upon some local descriptors recently proposed in the steganalysis field, which are selected and merged based on an ad hoc measure of reliability. The block-matching tool leverages on the patchmatch algorithm for fast search of candidate matchings. Both tools are fine-tuned so as to optimize their fusion which, in turn, exploits the respective strengths and weaknesses of each tool. The proposed technique ranked first in phase 1 of the first Image Forensics Challenge organized in 2013 by the IEEE Signal Processing Society.

We propose an image forgery localization technique which fuses the outputs of three complementary tools, based on sensor noise, machine-learning and block-matching, respectively. To apply the sensor noise tool, a preliminary camera identification phase was required, followed by estimation of the camera fingerprint, and then forgery detection and localization. The machine-learning is based on a suitable local descriptor, while block-matching relies on the PatchMatch algorithm. A decision fusion strategy is then implemented, based on suitable reliability indexes associated with the binary masks. The proposed technique ranked first in phase 2 of the first Image Forensics Challenge organized in 2013 by the IEEE Information Forensics and Security Technical Committee (IFS-TC).

In this work we propose a new algorithm for copy-move forgery detection and localization, based on the fast computation of a dense nearest-neighbor field. To this end, we use PatchMatch, an iterative randomized algorithm for nearest-neighbor search, which exploits the regularity of natural images to converge very rapidly to a near-optimal and smooth field. We modify the basic algorithm to gain robustness against rotations, while keeping the original computational efficiency. Experimental results show the proposed technique to outperform almost uniformly all tested reference techniques in terms of both accuracy and speed.

Digital camera identification is a very active research area, with important applications in the forensics field. Several approaches have been proposed in recent years for this task. One of the most promising is based on the estimation of the sensor noise pattern, used as a sort of camera fingerprint. However, a clever attacker can estimate a camera fingerprint and use it maliciously: this calls for new countermeasures, and so on, in a typical two-party game. In this paper we consider the triangle test, a well-know countermeasure against fake fingerprint attacks, and propose a new algorithm for improving the attacker's success rate. Numerical experiments show that, in typical scenarios, the proposed algorithm improves significantly the attacker performance.

Despeckling techniques based on the nonlocal approach provide an excellent performance, but exhibit also a remarkable complexity, unsuited to time-critical applications. In this letter, we propose a fast nonlocal despeckling filter. Starting from the recent SAR-BM3D algorithm, we propose to use a variable-size search area driven by the activity level of each patch, and a probabilistic early termination approach that exploits speckle statistics in order to speed up block matching. Finally, the use of look-up tables helps in further reducing the processing costs. The technique proposed conjugates excellent performance and low complexity, as demonstrated on both simulated and real-world SAR images and on a dedicated SAR despeckling benchmark.