News | IPLab - Image Processing Laboratory

IEEE Access

Wed, 12 Nov 2025 11:00:43 +0000

We are thrilled to announce that our paper

“Attractor-Aware Hyperbolic Lipschitz-Constrained Reinforcement Learning for FX Market: LLM-Structured Chain of Thought with Lyapunov–Entropy Dynamics”

has been accepted for publication in IEEE Access Journal (Q1)

Authors: Francesco Rundo, Ph.D., Simone Brancozzi, Massimo Orazio Spata, Ph.D., Michael S. Rundo, and Sebastiano Battiato.

-Innovation & Pipeline: this work introduces a fully explainable, geometry-aware deep learning pipeline for FX market modeling and trading, merging physics of chaos, non-Euclidean learning, and reinforcement control:

-LLM-Structured Chain-of-Thought (s-CoT): A novel reasoning framework where a Large Language Model builds structured, evidence-based s-CoT that connect heterogeneous signals (market data, technical indicators, macro events, CoT reports, Lyapunov and entropy diagnostics) into interpretable causal rules.

These reasoning graphs are then embedded in hyperbolic space, allowing the model to preserve hierarchical dependencies and express multi-scale relationships between causes, effects, and price dynamics, effectively transforming market reasoning into a geometric structure.

-Reinforcement Learning in Hyperbolic Manifolds: The RL policy evolves on curved space, where magnet prices act as dynamic attractors guiding stable, risk-aware actions.

-Lyapunov Exponents & Permutation Entropy: Measure local stability, predictability horizon, and structural disorder, turning chaotic-system diagnostics into quantitative market intelligence.

-Lipschitz-Constrained Learning: Ensures robustness and smooth adaptation under regime shifts and distributional changes, enabling continual learning in volatile financial environments.

The result is an interpretable reasoning-to-action pipeline, capable of producing daily trading decisions with explicit risk budgets, transparent structure, and resilience across market regimes.

Applications:

-Banking Institutions, designing risk-managed structured financial instruments.

-Investment Funds, constructing adaptive, regime-aware portfolio allocation strategies.

-Quantitative Traders & Analysts, understanding markets as nonlinear dynamical systems and enhancing decision timing and risk control.

This research stems from advanced deep learning R&D on non-conventional geometric spaces applied to financial systems, conducted at the Department of Mathematics and Computer Science of the University of Catania (AI4ILF Group@IPLAB).

#AI #ReinforcementLearning #HyperbolicGeometry #Lyapunov #DeepLearning #Finance #QuantitativeFinance #Research #IEEEAccess #UniversityOfCatania #AI4ILF #Innovation #IPLAB

Paper – Early Access Link:

https://ieeexplore.ieee.org/document/11215735?fbclid=IwY2xjawOBSBBleHRuA2FlbQIxMQBzcnRjBmFwcF9pZBAyMjIwMzkxNzg4MjAwODkyAAEe09EenR2obAGLfEG-0Dls2vDNH9Q-Ax5ceX-RBy7YE5dRBE9HUkuPhuiNTnM_aem_O8BprQV1Y5-hxubt0EmNgA

MetroXRAINE 2025

Wed, 12 Nov 2025 10:58:55 +0000

From October 22 to 24, Massimo Orazio Spata have attended MetroXRAINE 2025 IEEE Conference in Ancona, where I presented with an oral two research papers during the main session.

The conference provided an excellent opportunity to discuss recent advances in AI, metrology, and embedded systems. During the conference dinner I had the pleasure and honor to meet and exchange ideas in person, with the Director of China National Information Technology.

Presented papers:

Advanced Knowledge Distillation Pipeline with Dynamic Weighted-Smoothed Loss for a Robust Deep Network Deployment in Resource-Constrained Embedded Systems

Authors: Francesco Rundo (University of Catania, Italy); Giulia Castagnolo and Carmelo Pino (STMicroelectronics, Italy); Massimo Orazio Spata (University of Catania, Italy); Angelo Alberto Messina and Michele Calabretta (STMicroelectronics, Italy); Sebastiano Battiato (University of Catania, Italy)

Advanced Non-Local Pipeline for Domain-Adapted Intelligent Power Device Defects Analysis for a Robust Traction Inverter in Next Generation Electric Cars

Authors: Carmelo Pino (STMicroelectronics, Italy); Francesco Rundo (University of Catania, Italy); Giulia Castagnolo (STMicroelectronics, Italy); Massimo Orazio Spata (University of Catania, Italy); Angelo Alberto Messina and Michele Calabretta (STMicroelectronics, Italy); Sebastiano Battiato (University of Catania, Italy)

Thanks to my co-authors and to the MetroXRAINE 2025 organizers for a well-organized and stimulating event.

#MetroXRAINE2025 #AI #DeepLearning #KnowledgeDistillation #EmbeddedSystems #ElectricVehicles #Research

IEEE/CVF WACV (Winter Conference on Application of Computer Vision) 2026

Wed, 12 Nov 2025 10:57:39 +0000

Novel Metric for Detecting Memorization in Generative Models for Brain MRI Synthesis’ accepted at IEEE/CVF WACV (Winter Conference on Application of Computer Vision) 2026′

Co-authored by: Antonio Scardace, Lemuel Puglisi, Francesco Guarnera, Sebastiano Battiato, Daniele Ravì

https://github.com/brAIn-science/DeepSSIM

https://arxiv.org/abs/2509.16582

#medicalimaging

EURASIP Journal on Information Security

Wed, 12 Nov 2025 10:56:39 +0000

CNNMC: A Convolutional Neural Network with Monte Carlo Dropout for Speaker Recognition.

Authors: Massimo Orazio Spata, Alessandro Ortis, Georgia Fargetta, Sebastiano Battiato

Finally accepted at EURASIP Journal on Information Security (Q2, IF 2.1)

Pushing CNNs Beyond the Usual Limits!

We’re excited to share our latest work presenting a new CNN architecture with a custom Monte Carlo Dropout => CNNMC.

We’ve been experimenting with a new CNN using a new customized Monte Carlo Dropout. The CNNMC use this Dropout not only in the training phase, but also during inference.

Why does this matter?

This new CNNMC architecture:

– captures more information.

– reduces variance.

– makes decisions more robust and stable with unseen and noisy data.

The application: Speaker verification and Speaker identification.

The results: higher accuracy for identification task and lower EER for verification task, at inference time compared to other SOTA models: CNN, GMM, LDA, X-vector, I-vector, KNN.

We measured performances with 4 different datasets: BioVID, TIMIT, LibriSpeech, VoxCeleb.

Our Dataset: https://www.dmi.unict.it/spata/BiovidChallenge/

I’m sincerely grateful to all co-authors for the valuable collaboration and for the opportunity to work together on this research.

#SpeakerRecognition #DeepLearning #MonteCarloDropout #Biometrics #CNN #AIResearch #BioVid

AAAI 2026

Wed, 12 Nov 2025 10:55:55 +0000

I’m super excited to share that our paper “TRUST: Leveraging Text Robustness for Unsupervised Domain Adaptation” has been accepted at AAAI 2026!

TRUST introduces a new approach to unsupervised domain adaptation (UDA), designed to handle complex domain shifts by transferring the robustness of the language modality into the vision one.

TRUST generates pseudo-labels from image captions, estimates their uncertainty using CLIP image–text similarities, and aligns image-text representations using a novel multimodal contrastive loss.

I’m incredibly proud that our work was selected among 23,680 submissions, with an insane acceptance rate (17.60%).

A special thank all of my co-authors Sebastiano Battiato Valerio Giuffrida and Devis Tuia.

Simulation Modelling Practice and Theory

Wed, 12 Nov 2025 10:55:25 +0000

I’m very pleased to share that our latest paper, “Multi-agent reinforcement learning and variational inequality models for international trade networks under crisis”, has just been published in Simulation Modelling Practice and Theory (Elsevier, Q1 in Computer Science). https://doi.org/10.1016/j.simpat.2025.103219

This work, co-authored with Laura De Natale, Laura R.M. Scrimali, and [Sebastiano Battiato](https://www.linkedin.com/home?originalSubdomain=it#), introduces a hybrid Multi-Agent Reinforcement Learning (MARL) framework for solving variational inequality models in multi-commodity trade networks.

Our method bridges the mathematical rigor of equilibrium theory with the adaptive learning capabilities of MARL, addressing complex global trade dynamics, especially under crisis and disruption scenarios affecting food security and supply chains.

We demonstrate how the integration of Gradient-based Learning Rate scheduling (GLR), adaptive exploration, and dual reward mechanisms enables faster convergence and improved stability compared to MAPPO and MADDPG baselines.

Grateful to all co-authors and collaborators for their valuable contributions and discussions that made this work possible.

https://www.sciencedirect.com/science/article/pii/S1569190X25001546?via%3Dihub

1st AADD‑2025 Challenge: Adversarial Attacks on Deepfake Detectors

Wed, 12 Nov 2025 10:53:47 +0000

We are very proud to announce that the 1st AADD‑2025 Challenge: Adversarial Attacks on Deepfake Detectors, held as a grand challenge at ACM Multimedia 2025, was a remarkable success.!! The challenge brought together teams from around the world to address a critical and timely issue: how synthetic media detection systems can be compromised by adversarial manipulations.

With rigorous evaluation, a rich dataset of GAN- and diffusion-based deepfake images in both high- and low-quality resolutions, and a leaderboard revealing outstanding contributions, the event provided a powerful forum for advancing the state of the art in adversarial resilience for deepfake detectors.

We extend our deepest thanks to the organisers of the challenge:

Organizers

Luca Guarnera, Research Fellow, Department of Mathematics & Computer Science, University of Catania, Italy

Francesco Guarnera, Research Fellow, Department of Mathematics & Computer Science, University of Catania, Italy

Sebastiano Battiato, Full Professor, Department of Mathematics & Computer Science, University of Catania, Italy

Giovanni Puglisi, Associate Professor, Department of Mathematics & Computer Science, University of Cagliari, Italy

Zahid Akhtar, Associate Professor, State University of New York Polytechnic Institute, USA

Mirko Casu, PhD Student, Department of Mathematics & Computer Science, University of Catania, Italy

Orazio Pontorno, PhD Student, Department of Mathematics & Computer Science, University of Catania, Italy

Claudio Vittorio Ragaglia, PhD Student, Department of Mathematics & Computer Science, University of Catania, Italy

A special thanks to all participating teams.

#Deepfake #AdversarialDeepfake #acmmultimedia2025 #IPLab

1st Deepfake Forensics Workshop: Detection, Attribution, Recognition, and Adversarial Challenges in the Era of AI-Generated Media

Wed, 12 Nov 2025 10:53:10 +0000

We are thrilled to announce that the 1st Deepfake Forensics Workshop: Detection, Attribution, Recognition, and Adversarial Challenges in the Era of AI-Generated Media, held as part of ACM Multimedia 2025, was a resounding success!
https://iplab.dmi.unict.it/mfs/acm-dff-ws-2025/

With high-quality keynote talks and a diverse program of accepted papers, the event provided a rich forum to explore the state-of-the-art and emerging challenges in multimedia forensics. Topics ranged from multimodal deepfake detection, model attribution, to adversarial forensics, interpretability, dataset bias and the legal & ethical implications of AI-generated content.

We express our sincere thanks to the Organising Committee for their outstanding work:

General Chairs

Luca Guarnera (Research Fellow, Department of Mathematics & Computer Science, University of Catania)

Francesco Guarnera (Research Fellow, Department of Mathematics & Computer Science, University of Catania)

Workshop Co-Chairs

Sebastiano Battiato (Full Professor, Department of Mathematics & Computer Science, University of Catania)

Giovanni Puglisi (Associate Professor, Department of Mathematics & Computer Science, University of Cagliari)

Zahid Akhtar (Associate Professor, State University of New York Polytechnic Institute)

Track Co-Chairs

Mirko Casu (PhD Student, Department of Mathematics & Computer Science, University of Catania)

Orazio Pontorno (PhD Student, Department of Mathematics & Computer Science, University of Catania)

Claudio Vittorio Ragaglia (PhD Student, Department of Mathematics & Computer Science, University of Catania)

Thank you to all authors, reviewers and attendees for making this workshop such a memorable and impactful event.

#DFF-2025 #Deepfake #acmmultimedia2025 #IPLab

Ecological Informatics

Wed, 12 Nov 2025 10:52:15 +0000

Georgia Fargetta, Stefano Anile, Luigi Priano, Massimo Orazio Spata, Alessandro Ortis, and Sebastiano Battiato are pleased to announce the publication of our article:

Bridging AI and wildlife conservation: Classifying wildcats from genetically validated images in Ecological Informatics (Elsevier, Q1 – Computer Science).
https://www.sciencedirect.com/science/article/pii/S1574954125004765

This work introduces an AI-based recognition system designed to support wildlife researchers and conservationists in the classification of wild-living cats, including European wildcats (Felis silvestris), hybrids (F. silvestris × catus), and domestic cats (F. catus), using genetically validated real-world images.

A key outcome of the study is the development of a mobile application for automatic recognition, optimized for offline field use, enabling researchers and citizen scientists to classify cat images directly on mobile devices, even in remote environments.

All resources — including the dataset, model weights, and mobile recognition app — are available upon request at the following link: https://www.dmi.unict.it/ortis/wildcat/

This research represents a successful collaboration between computer scientists, zoologists, and citizen scientists, coordinated within the EuroWildcat network, and demonstrates how artificial intelligence can effectively contribute to ecological monitoring and biodiversity conservation.

A sincere thank you to all co-authors and to the EuroWildcat research group for their essential contribution and data sharing.

Plasmare il futuro con l’AI applicata

Wed, 12 Nov 2025 10:51:24 +0000

Happy to contribute next week (28/10/2025) – thanks to Xenia Progetti for inviting me!

Prof Giovanni Maria Farinella

IEEE Access

Wed, 12 Nov 2025 10:49:55 +0000

The paper entitled “A Novel Adversarial Gray-Box Attack on DCT-based Face Deepfake Detectors” has been accepted at IEEE Access

This work introduces a novel adversarial gray-box attack specifically designed to bypass deepfake detectors based on Discrete Cosine Transform (DCT) features. The proposed method manipulates the AC coefficient statistics of synthetic images so that they closely resemble those of real ones, while preserving high visual quality. The attack assumes full knowledge of the DCT feature extraction process but no access to the internal parameters of the classifiers, making it a gray-box scenario. Experimental results show that the attack is effective against detectors tested on deepfakes generated by both GANs and Diffusion Models, significantly reducing their reliability and exposing critical vulnerabilities in detection systems often considered interpretable and robust. This research highlights important implications for multimedia forensics and cybersecurity, raising awareness of the need for more resilient defense strategies against adversarial manipulations.

Authors: Francesco Guarnera, Luca Guarnera, Alessandro Ortis, Sebastiano Battiato, Giovanni Puglisi.

Link: https://ieeexplore.ieee.org/document/11192411

IEEE Access

Wed, 12 Nov 2025 10:49:12 +0000

The paper entitled “SignForensics: A Robust Framework for Forensic Offline Signature Verification with Enhanced Detection, Noise Removal, and Multi-Stage Authentication” has been accepted at IEEE Access

This work presents SignForensics, an innovative framework for forensic offline signature verification. The proposed pipeline integrates YOLOv10 for automatic signature detection in documents, CycleGAN for noise removal such as lines, stamps, and printed text, and customized deep learning models based on SigNet and Capsule Networks for advanced feature extraction. The approach adopts a multi-stage authentication strategy, first assessing the compatibility of a signature with a set of potential signers and then verifying its authenticity. The study provides a significant contribution to the field of multimedia forensics, offering a robust and reliable tool for signature authentication in real-world, noisy, and complex scenarios.

Authors: Giovanni Piccinini, Luca Guarnera, Sebastiano Battiato

Link: https://ieeexplore.ieee.org/document/11192260

IEEE Open Journal of the Computer Society

Wed, 12 Nov 2025 10:47:56 +0000

The paper “Hyperbolic Deep Learning System for Intelligent Gate-Driving and Health Monitoring of SiC MOSFETs”

has been accepted for publication in the IEEE Open Journal of the Computer Society (Impact Factor 8.2 — Q1 in Computer Science (misc)).

Our contribution was driven by a central key-technical pillar: Hyperbolic deep learning combined with Lipschitz-controlled layers. This approach allows us to build geometry-aware, stable, and continually adaptable models capable of learning even under domain shifts and evolving operating conditions.

We applied this novel backbone to the world of power electronics: Physics-aware modeling and intelligent health monitoring of power devices. By analyzing standard gate-driving data, our system can track device degradation and support reliable operation of Silicon/Silicon-Carbide MOSFETs, while remaining light-weight enough for real-time, embedded deployment.

Authors: Francesco Rundo, Ph.D., Angelo Alberto Messina, Michele Fiore, Michele Calabretta, Pasquale Coscia, Sebastiano Battiato

#UniCT – #STMicroelectronics · #UniMI – #IPLAB

Read the IEEE OJCS – Early Access version here: https://lnkd.in/d9-qyynR

This work is part of ongoing research led by AI4Industry@IPLAB, University of Catania, in partnership with industrial players such as STMicroelectronics, where we push the frontiers of deep learning on un-conventional spaces and bio-inspired models for next-generation industrial applications.

Stay tuned!

International Journal of Computer Vision

Wed, 12 Nov 2025 10:47:09 +0000

Good news!

The paper with title “Exocentric-to-Egocentric Adaptation for Temporal Action Segmentation with Unlabeled Synchronized Video Pairs” has been accepted for publication in the International Journal of Computer Vision (Q1, IF 9.3 in 2024)

Authors: Camillo Quattrocchi, Antonino Furnari, Daniele Di Mauro, Mario Valerio Giuffrida, Giovanni Maria Farinella

Journal of Computer Vision

Wed, 12 Nov 2025 10:46:17 +0000

We are pleased to share that our paper “Uncertainty-guided Open-Set Source-Free Unsupervised Domain Adaptation with Target-private Class Segregation” – Mattia Litrico, Davide Talon, Sebastiano Battiato , Alessio Del Bue, Mario Valerio Giuffrida, Pietro Morerio has been accepted for publication at the International Journal of Computer Vision (IF: 9.3).

This is a big milestone and it is the result of a great collaboration between our group, the IIT and the University of Nottingham! This is a demonstration that sharing knowledge between research groups allows to achieve incredible results.

A big thank to all my co-authors for their effort. It was a long and inspiring journey!

ICIAP 2025

Wed, 12 Nov 2025 10:45:21 +0000

Congrats to our students Giuseppe Lando and Rosario Forte for the ICIAP 2025 Best Student Paper Award!

Human Behavior and Emerging Technologies

Wed, 12 Nov 2025 10:43:50 +0000

Can people really tell AI images from real ones?

Our new study, “A (Mid)Journey Through Reality”, published today (!) in Human Behavior and Emerging Technologies (Q1 in PSY/CS), reveals that humans perform at near-chance levels (47-55%) when trying to distinguish state-of-the-art AI-generated images from real photographs.

Also, we’re more likely to doubt AI-generated images, even when we’re wrong (impostor bias); we frequently change our minds to agree with an algorithm’s suggestion, even when it’s incorrect (automation bias), and men were significantly more susceptible to this latter across all experiments.

Read the full paper here: https://doi.org/10.1155/hbe2/9977058

Luca Guarnera, Ignazio Zangara, Pasquale Caponnetto, Sebastiano Battiato

WACV 2026

Wed, 12 Nov 2025 10:43:13 +0000

Good news!

Three papers accepted at WACV 2026! (iCore rank: A)

ProSkill: Segment-Level Skill Assessment in Procedural Videos (in collaboration with Toyota – cofirst authors: Michele Mazzamuto & Di Daniele Di Mauro

Ego-EXTRA: video-language Egocentric Dataset for EXpert-TRAinee assistance (in collaboration with META – cofirst author Francesco Ragusa & Michele Mazzamuto)

Online Episodic Memory Visual Query Localization with Egocentric Streaming Object Memory (in collaboration with UniUD – outcome of PNRR PRIN Project UniCT-UniUD led by AF for UniCT)

WACV received 1,329 valid paper submissions in Round 1, excluding papers that were desk rejected or withdrawn. Of these, 85 were accepted, 507 were rejected, and 732 were invited to resubmit to Round 2.

Acceptance rate: 6.4%

Human Behavior and Emerging Technologies

Wed, 12 Nov 2025 10:42:17 +0000

We’re thrilled to share that our manuscript “A (Mid)Journey Through Reality: Assessing Accuracy, Impostor Bias, and Automation Bias in Human Detection of AI-Generated Images” has been accepted for publication in Human Behavior and Emerging Technologies (IF: 3; AR: 16%; Q1 in Psychology, 96th p.; Q1 in Computer Science, 88th p.).

This interdisciplinary project bridges psychology and computer science to explore how (and if…) people detect AI-generated images, with important implications for media literacy, policy, and AI safety. Specifically, it builds on and evolves our earlier work “GenAI Mirage: The Impostor Bias and the Deepfake Detection Challenge” (https://doi.org/10.1016/j.fsidi.2024.301795), extending those findings with a larger empirical study that was primarily designed to validate the impostor bias, the tendency for people to systematically distrust (or misattribute) the authenticity of images in the age of generative AI.

(As such, this paper both deepens and empirically tests the hypotheses we proposed in the first study.)

We began structuring this research almost two years ago, and I’m very delighted to make it finally see the light.

Last but not least, a big thank you to all the participants!

Authors: Mirko Casu, Luca Guarnera, Ignazio Zangara, Pasquale Caponnetto, Sebastiano Battiato.

DFF-2025 at ACM Multimedia!

Wed, 12 Nov 2025 10:41:10 +0000

Good news!

Excited to share that our paper, “What if Retrieval Could Work Before Decoding? The Case of JPEG AI Latents for Deepfake Source Attribution,” has been accepted to DFF-2025 at ACM Multimedia!

We show how latent tensors from JPEG AI can be directly used to discriminate image semantics and perform deepfake source attribution, without any training, just image-to-image retrieval. Results are promising and open up new directions for image forensics.

Thanks to my coauthors Lorenzo Catania, Francesco Guarnera, Dario Allegra, Prof Sebastiano Battiato

ICCV 2025 (PHAROS Workshop)

Wed, 12 Nov 2025 10:39:46 +0000

We are thrilled to announce that our paper titled “𝗗𝗲𝗲𝗽𝗦𝗲𝗴-𝗠𝗦: 𝗔 𝟯𝗗 𝗻𝗲𝘁𝘄𝗼𝗿𝗸 𝗯𝗮𝘀𝗲𝗱 𝗼𝗻 𝗵𝗶𝗲𝗿𝗮𝗿𝗰𝗵𝗶𝗰𝗮𝗹 𝗺𝘂𝗹𝘁𝗶-𝘀𝗰𝗮𝗹𝗲 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗳𝗼𝗿 𝗠𝗥𝗜 𝗺𝘂𝗹𝘁𝗶𝗽𝗹𝗲 𝘀𝗰𝗹𝗲𝗿𝗼𝘀𝗶𝘀 𝗹𝗲𝘀𝗶𝗼𝗻 𝘀𝗲𝗴𝗺𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻” has been accepted at the “𝘗𝘏𝘈𝘙𝘖𝘚 – 𝘈𝘥𝘢𝘱𝘵𝘢𝘵𝘪𝘰𝘯, 𝘍𝘢𝘪𝘳𝘯𝘦𝘴𝘴, 𝘌𝘹𝘱𝘭𝘢𝘪𝘯𝘢𝘣𝘪𝘭𝘪𝘵𝘺 𝘪𝘯 𝘈𝘐 𝘔𝘦𝘥𝘪𝘤𝘢𝘭 𝘐𝘮𝘢𝘨𝘪𝘯𝘨” Workshop, in conjunction with ICCV 2025 (Honolulu, Hawaii)

This work presents 𝗗𝗲𝗲𝗽𝗦𝗲𝗴-𝗠𝗦, a novel 3D deep learning architecture tailored for accurate and robust segmentation of Multiple Sclerosis lesions from multimodal MRI.

Our model introduces three key modules:

𝘚𝘲𝘶𝘦𝘦𝘻𝘦-𝘢𝘯𝘥-𝘈𝘵𝘵𝘦𝘯𝘵𝘪𝘰𝘯 𝘣𝘭𝘰𝘤𝘬𝘴 — to enhance focus on relevant lesion regions

𝘋𝘦𝘦𝘱 𝘚𝘶𝘱𝘦𝘳𝘷𝘪𝘴𝘪𝘰𝘯 — for effective hierarchical learning

𝘈𝘵𝘳𝘰𝘶𝘴 𝘚𝘱𝘢𝘵𝘪𝘢𝘭 𝘗𝘺𝘳𝘢𝘮𝘪𝘥 𝘗𝘰𝘰𝘭𝘪𝘯𝘨 — to capture multi-scale contextual features through parallel dilated convolutions

By addressing the spatial complexity and heterogeneity of MS lesions in 3D volumes, 𝗗𝗲𝗲𝗽𝗦𝗲𝗴-𝗠𝗦 pushes forward the capabilities of automated neuroimaging-based analysis.

𝗔𝘂𝘁𝗵𝗼𝗿𝘀:

Alessia Rondinella

Francesco Guarnera

Francesco Rundo

Sebastiano Battiato

𝗦𝘁𝗮𝘆 𝘁𝘂𝗻𝗲𝗱 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗼𝗳𝗳𝗶𝗰𝗶𝗮𝗹 𝗽𝗿𝗲𝗽𝗿𝗶𝗻𝘁 𝗮𝗻𝗱 𝗰𝗼𝗱𝗲 𝗿𝗲𝗹𝗲𝗮𝘀𝗲!

#DeepSegMS #ICCV2025 #PHAROS #MedicalImaging #MultipleSclerosis #3DSegmentation #Neuroimaging #DeepLearning

ICCV 2025 (PHAROS Workshop)

Wed, 12 Nov 2025 10:38:56 +0000

We are proud to announce that our paper titled “Dopamine-Inspired Neuro-Modulated Radiomic Pipeline for Computed-Tomography-Driven Prediction of Immunotherapy Response in Metastatic Cancer Treatment” has been accepted at the “PHAROS: Adaptation, Fairness, Explainability in AI Medical Imaging” Workshop in conjuction with the International Conference on Computer Vision (ICCV 2025).

Authors:

Francesco Rundo, Ph.D. University of Catania

Massimo Orazio Spata University of Catania

Carmelo Pino STMicroelectronics

Sebastiano Battiato University of Catania

This work introduces a novel biologically-inspired computational framework that integrates dopamine-inspired neuro-modulation mechanisms into ad-hoc hybrid spatio-temporal augmented deep pipeline for enhanced prediction of PD-L1 inhibitors immunotherapy response from patients CT scans imaging. This work introduces a novel bio-inspired mechanism that emulates dopaminergic action in the human brain, to address the “stability-plasticity” dilemma in deep learning systems.

We believe this interdisciplinary contribution bridges neuroscience, medical imaging, and AI-driven oncology.

We look forward to sharing our findings and engaging with the research community in Honolulu Hawaii this October.

Stay tuned for more details and the official preprint release.

#PHAROS #ICCV2025 #Radiomics #NeuroModulation #MedicalImaging #DeepLearning #Immunotherapy #ComputationalOncology

ACM Multimedia 2025

Wed, 12 Nov 2025 10:38:16 +0000

We are excited to share that our paper

“Towards Reliable Audio Deepfake Attribution and Model Recognition: A Multi-Level Autoencoder-Based Framework”

has been accepted for presentation at the DFF Workshop @ ACM Multimedia 2025, to be held in Dublin !

In this work, we present LAVA, a multi-level architecture for audio deepfake attribution that combines a convolutional autoencoder, attention mechanisms, and open-set rejection to recognize both the technology and the specific model used to synthesize fake audio.

The framework is trained and evaluated across multiple datasets (CodecFake, ASVspoof2021, FakeOrReal) and demonstrates strong performance even under open-set conditions and distribution shifts (ASVspoof2019 LA).

Authors:

Check out the architecture below!

UniCt ranked as 4th among Italian Universities for the area of “Computer Vision”

Wed, 12 Nov 2025 10:37:35 +0000

Computer Science Ranking: Happy to see UniCt ranked as 4th among Italian Universities for the area of “Computer Vision”

https://csrankings.org/#/index?vision&itto

ICVSS 2025

Wed, 12 Nov 2025 10:36:47 +0000

Multimedia Tools and Applications (Springer)

Wed, 12 Nov 2025 10:36:08 +0000

Our paper “DeepFeatureX-SN: Generalization of Deepfake Detection via Contrastive Learning” has been accepted for publication in Multimedia Tools and Applications (Springer)!

In this work, we introduce DeepFeatureX-SN, a novel deep learning model based on Siamese networks and contrastive learning designed to detect both GAN and Diffusion-based synthetic images. The architecture leverages three specialized feature extractors (for real, GAN, and DM-generated images), whose outputs are fused into a robust CNN classifier.

Highlights:

State-of-the-art performance on generalization tasks involving unseen generative models.

Strong robustness to image manipulations (JPEG compression, rotation, resizing).

Outperforms existing deepfake detection methods across multiple benchmarks.

Authors: Orazio Pontorno, Luca Guarnera, Sebastiano Battiato

Code and dataset will be released soon at: https://github.com/mfs-iplab/DeepFeatureX-SN

#DeepfakeDetection #AIForensics #DeepLearning #GAN #DiffusionModels #DeepFeatureX #IPLab

ICVSS 2025

Wed, 12 Nov 2025 10:34:18 +0000

https://www.unictmagazine.unict.it/alla-icvss-si-studia-la-computer-vision-spatial-ai?fbclid=IwY2xjawOBQhJleHRuA2FlbQIxMQBzcnRjBmFwcF9pZBAyMjIwMzkxNzg4MjAwODkyAAEetqKImtm0dhXE1F2cd4A_8lSt7CMkuxrkAK7aEJiOumFOfj4pyBezUle3M-U_aem_uioBtjiGTsA3qWakQv852A

ICIAP 2025

Wed, 12 Nov 2025 10:33:32 +0000

Good news!

“A Real-Time System for Egocentric Hand-Object Interaction Detection in Industrial Domains” accepted at the ICIAP Workshop on Human-Object Interaction: Egocentric & Exocentric Perspectives

Congrats to our master sudents Alessandro Catinello and Antonio Finocchiaro!

MetroAUTOMOTIVE 2025

Wed, 12 Nov 2025 10:32:37 +0000

Last week, Massimo Orazio Spata had the pleasure of attending IEEE Metroautomotive 2025 conference in Parma, where I had the opportunity to present the latest research developed within the joint Lab DMI-STMicroelectronics.

Our work, titled:

“Silicon-Carbide Power Device X-Ray Screening via Attention-Based Deep Network for a Robust Traction-Inverter in Electric Vehicles” was the result of a fruitful collaboration between academia and industry.

Authors:

Francesco Rundo (University of Catania, Italy)

Giulia Castagnolo (STMicroelectronics, Italy)

Carmelo Pino (STMicroelectronics, Italy)

Massimo Orazio Spata (University of Catania)

Angelo Alberto Messina (STMicroelectronics, Italy)

Sebastiano Battiato (University of Catania, Italy)

Moreover 2 papers accepted for IEEE MetroXRAINE 2025 conference:

-Advanced Non-Local Pipeline for Domain-Adapted Intelligent Power Device Defects Analysis for a Robust Traction Inverter in Next Generation Electric Cars, Francesco Rundo, Carmelo Pino, Giulia Castagnolo, Massimo Orazio Spata, Angelo A. Messina, Michele Calabretta, Sebastiano Battiato.

-Advanced Knowledge Distillation Pipeline with Dynamic Weighted-Smoothed Loss for a Robust Deep Network Deployment in Resource-Constrained Embedded Systems, Francesco Rundo, Carmelo Pino, Giulia Castagnolo, Massimo Orazio Spata, Angelo A. Messina, Michele Calabretta, Sebastiano Battiato.

A big thank to our Academic-Industrial Conjunction Team to University of Catania, to STMicroelectronics, to the organizers and to the research community for the engaging discussions and valuable feedback!

#Metroautomotive2025 #DeepLearning #XRayImaging #SiC #TractionInverter #ElectricVehicles #AcademicIndustryCollaboration #STMicroelectronics #DMI #CorsoDiLaureaInformatica #Unict #AI4EV

MetroXRAINE 2025

Wed, 12 Nov 2025 10:31:39 +0000

I am pleased to announce that two papers have been accepted at IEEE MetroXRAINE 2025:

1. Deepfake Forensic Analysis: Source Dataset Attribution and Legal Implications of Synthetic Media Manipulation. Authors: Massimiliano Cassia, Luca Guarnera, Mirko Casu, Ignazio Zangara, Sebastiano Battiato
This work presents a forensic framework to attribute the source dataset (e.g., CelebA or FFHQ) of deepfake images generated by GANs, using frequency analysis (DCT/FFT), color histograms, and local features (SIFT). The system achieves up to 95.5% classification accuracy and addresses legal implications related to copyright and privacy.
Congratulations to Massimiliano on his first paper

2. Omnidirectional Photographic Capture for VR-Based Representation of Indoor Space: a Subjective Study. Authors: Giovanni Lo Monaco, Luca Guarnera, Sebastiano Battiato, Sergio Palazzo, Salvatore Livatino
This study presents a lightweight approach for immersive VR using 360-degree imagery to represent real indoor environments. A real room was captured with high-resolution omnidirectional photography and mapped onto a virtual sphere. Through a user study with 22 participants, the method was evaluated in terms of presence, depth perception, visual realism, and distortion. Results show that while some visual artifacts exist, the technique provides strong visual immersion and spatial comprehension, making it suitable for rapid VR deployment in education, training, and documentation.

#IPLab #MultimediaForensics #VR

IJCNN conference

Wed, 12 Nov 2025 10:30:36 +0000

Two papers were presented at VERIMEDIA: International Workshop on Media Verification and Integrity – IJCNN conference. Official website https://sites.unica.it/ff4all/verimediaijcnn/

WILD: a new in-the-Wild Image Linkage Dataset for synthetic image attribution. Authors: Bongini, Pietro; Mandelli, Sara; Montibeller, Andrea; Casu, Mirko; Pontorno, Orazio; Ragaglia, Claudio Vittorio; Zanchetta, Luca; Aquilina, Mattia; Majid Wani, Taiba; Guarnera, Luca; Tondi, Benedetta; Boato, Giulia; Bestagini, Paolo; Amerini, Irene; De Natale, Francesco; Battiato, Sebastiano; Barni, Mauro.

End-to-end audio deepfake detection from raw waveforms: a RawNet-based approach with cross-dataset evaluation. Authors: Di Pierno, Andrea; Guarnera, Luca; Allegra, Dario; Battiato, Sebastiano

#deepfake #syntheticcontent #IPLab #ff4ll

Premio Archimede 2025

Wed, 12 Nov 2025 10:29:55 +0000

Congratulazioni ad Alfio e Riccardo nostri studenti!

https://www.facebook.com/share/1JAzmhq73Q/

ECCV 2026

Wed, 12 Nov 2025 10:28:42 +0000

ECCV 2026 will be in Malmo, Sweden

Sept 8-13, 2026

Happy to be part of the Organizing Committee!

Stay tuned!

https://eccv.ecva.net

Ital-IA 2025

Wed, 12 Nov 2025 10:28:07 +0000

We are going to run the workshop on “AI for Industry” in Trieste next week as part of the “Italian Conference CINI on Artificial Intelligence (Ital-IA)”

List of contributions at the following link:

https://www.ital-ia2025.it/workshop/ai-per-l-industria.html

ICVSS 2025 – Poster Session

Wed, 12 Nov 2025 10:27:31 +0000

ICVSS 2025 – Poster Session

List of posters accepted at ICVSS 2025 is online!

https://icvss.dmi.unict.it/icvss2025/Posters

See booklet of abstracts at the following link:

https://icvss.dmi.unict.it/…/icvss_poster_booklet.pdf

ICIAP 2025

Wed, 12 Nov 2025 10:27:01 +0000

Awesome! 5 papers accepted at ICIAP 2025

Great performances of our master students (all first authors) Alessandro Sebastiano Catinello, Antonio Finocchiaro, Giuseppe Lando

Alessandro Sebastiano Catinello Et al.: Mamba-OTR: a Mamba-based Solution for Online Take and Release Detection from Untrimmed Egocentric Video

Antonio Finocchiaro Et al.: Efficient Calisthenics Skills Classification through Foreground Instance Selection and Depth Estimation

Zaira Manigrasso Et al. (In collaboration with University of Udine): T-EVO: Tracking in Egovision for Online Visual Episodic Memory

Alessandro Sebastiano Catinello et al. (In collaboration with Universities of Udine and Macerata): Ego and exo views for an object-level human behavior analysis and understanding through tracking in retail spaces

Giuseppe Lando et al.: “How Far Can Off-the-Shelf Multimodal Large Language Models Go in Online Episodic Memory Question Answering?”

We Make Future 2025

Wed, 12 Nov 2025 10:26:17 +0000

Grazie a quanti hanno rappresentato IPLAB alla scorsa edizione di “We Make Future” di Bologna

CVPR25

Wed, 12 Nov 2025 10:25:25 +0000

CVPR25, here we come!

I’m excited to attend the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR 2025) in Nashville and to be part of the Women in Computer Vision (WiCV) Workshop!

You can find me at the workshop the 12th of June (as in the pic ), where I’ll be presenting two posters:

“Exploiting Adversarial Learning and Topology Augmentation for Open-Set Visual Recognition” – full paper, in collaboration with Rosa Zuccarà, Alessandro Ortis, and Sebastiano Battiato.

“Explainable AI for Enhancing Image Transformations in Cellular Neural Networks” – extended abstract, with Francesco Rundo and Sebastiano Battiato.

CVPR25!

Wed, 12 Nov 2025 10:24:44 +0000

On our way to Nashville for CVPR25!

We will be presenting “Gazing Into Missteps: Leveraging Eye-Gaze for Unsupervised Mistake Detection in Egocentric Videos of Skilled Human Activities”, a paper in collaboration with University of Tokio which was accepted at the main conference! (Acceptance rate 22.1% – 2878 papers accepted out of 13008 valid submissions)

See preprint here: https://arxiv.org/pdf/2406.08379.pdf

Also, we are going to disseminate to different workshops

See our aggenda in pictures

Stay tuned for news!

We Make Future 2025

Wed, 12 Nov 2025 10:23:47 +0000

At We Make Future 2025, we’re presenting our research in various fields.

I’m showcasing an AI-driven smoking detection system that uses wearables, as well as a study on impostor bias in multimedia forensics.

Claudio Ragaglia is presenting a patented 3D VR Ballistic Analysis System with iCTLab and early findings from HPC Spoke 1 in multimedia forensics.

Massimo Orazio Spata is introducing a patent-pending smartphone-based fluorescent sensor for detecting plant pathogens.

Thanks to the University of Catania, Fondazione SAMOTHRACE, ICSC HPC, and Prof. Sebastiano Battiato for the support.

📢 New Dataset Release!

Wed, 12 Nov 2025 10:22:37 +0000

New Dataset Release!

The MSLesSeg dataset for Multiple Sclerosis Lesion Segmentation is now publicly available — and we’re thrilled to announce its official publication in Nature Scientific Data!

MSLesSeg marks a milestone in MS research: it is the largest expert-annotated dataset for MS lesion segmentation currently available at the state of the art.

The dataset includes multimodal MRI scans acquired from patients with MS, enriched with expert-validated lesion masks, comprehensive clinical data, and benchmark comparisons against cutting-edge segmentation algorithms.

What can you use MSLesSeg for?

Developing AI-based segmentation algorithms

Advancing clinical and translational neuroimaging research

Evaluating and comparing automated methods in realistic scenarios

A portion of MSLesSeg was successfully used in the ICPR 2024 MSLesSeg Challenge, which gathered strong international participation and confirmed the dataset’s value for benchmarking AI approaches in medical imaging.

Read the full article: https://www.nature.com/articles/s41597-025-05250-y

Download the dataset: https://doi.org/10.6084/m9.figshare.27919209

Explore more from our research: https://iplab.dmi.unict.it/medicalimaging/

A huge thank you to all co-authors for their outstanding contributions!

Francesco Guarnera

Alessia Rondinella

Elena Crispino

Giulia Russo

Clara Di Lorenzo

Davide Maimone

Francesco Pappalardo

Sebastiano Battiato

#MultipleSclerosis #Neuroimaging #AIinHealthcare #MedicalImaging #DeepLearning #BrainLesions #ScientificData #DatasetRelease

The Visual Computer

Wed, 12 Nov 2025 10:21:52 +0000

The paper “Benchmarking Computer Vision Architectures for Cloud Detection from Lidar Ceilometer Backscatter Data”

has been accepted for publication in The Visual Computer!

Abstract: Cloud detection is fundamental for accurate weather monitoring, often achieved through remote sensing technology, such as satellite imagery or radar. This study explores the use of lidar ceilometer backscatter data, a rich but noisy source of atmospheric information, to enhance cloud detection. Leveraging data acquired from a Lufft CHM 15k ceilometer over three months near Mount Etna, Italy, we gathered a novel dataset comprising time-height plots derived from backscatter profiles. The Weather Research and Forecasting (WRF) model was used for ground-truth data labeling, ensuring reliable model validation. We benchmarked state-of-the-art deep learning architectures, including CNN-based models (e.g., ResNet50, VGG16, InceptionV3, EfficientNet) and the Vision Transformer (ViT),

on our collected dataset. Among these, ResNet50 achieved the highest accuracy (89.57%), closely followed by ViT (89.36%), showcasing the efficacy of residual learning and transformer-based approaches in extracting complex patterns from atmospheric data. Our results highlight the potential of lidar-based systems for accurate cloud detection, complementing other remote sensing technologies. Our work contributes to the field by introducing a publicly available dataset and providing comprehensive benchmarking results that establish a baseline for future research. This study also opens avenues for broader applications of ceilometer data, such as the detection of pollutants and other atmospheric phenomena. Our

dataset is publicly available at https://zenodo.org/records/10616434

Authors: Alessio Chisari, Luca Guarnera, Alessandro Ortis, Wladimiro Carlo Patatu, Sebastiano Battiato, Valerio Giuffrida

#IPLab #cloudDetection

IJCNN 2025 Main conference Rome

Wed, 12 Nov 2025 10:21:07 +0000

‘Attention-Enhanced Convolutional Deep Architecture with Adaptive Jacobian Regularization for Joint Multi-Modal Optical and X-Ray Screening of Silicon-Carbide Power Devices’ accepted to IJCNN 2025 Main conference Rome

International Joint Conference on Neural Networks (IJCNN 2025) – VERIMEDIA Workshop

Wed, 12 Nov 2025 10:20:03 +0000

The paper WILD: a new in-the-Wild Image Linkage Dataset for synthetic image attribution has been accepted for presentation at 2025 International Joint Conference on Neural Networks (IJCNN 2025) – VERIMEDIA Workshop!(https://sites.unica.it/ff4all/verimediaijcnn/)

The paper introduces WILD, a new dataset designed for synthetic image source attribution, addressing the challenge of identifying which generative model created a given synthetic image. The dataset is divided into two parts:

Closed set: Contains 10,000 images from 10 popular commercial and open-source text-to-image generators (e.g., Adobe Firefly, DALL-E 3, Midjourney), with 1,000 images per generator. Images are generated using fixed, randomized prompts to ensure diversity and minimize bias.

Open set: Includes 10,000 images from 10 additional generators (e.g., StyleGAN variants, DALL-E 1) to simulate real-world scenarios where unknown models may be encountered.

Half of the images in both sets are post-processed (e.g., compressed, cropped, resized) to test model robustness. We evaluated seven baseline methods (e.g., CLIP-based classifiers, CNNs, Vision Transformers) on WILD. Key findings:

CNNs excel on plain images but degrade with post-processing.

Vision Transformers (ViT) show greater robustness to post-processing.

Open-set attribution remains challenging, with performance dropping significantly for post-processed images.

Thanks to all the forensic researchers from various Italian institutions who contributed to this study!

IPLAB Authors: Mirko Casu, Orazio Pontorno, Claudio Ragaglia, Luca Guarnera, Sebastiano Battiato

#Deepfake #DeepfakeAttribution #FF4LL #IPLab

International Joint Conference on Neural Networks (IJCNN 2025) – VERIMEDIA Workshop

Wed, 12 Nov 2025 10:19:23 +0000

We are pleased to inform you that the paper End-to-end Audio Deepfake Detection from RAW Waveforms: a RawNet-Based Approach with Cross-Dataset Evaluation has been accepted for presentation at 2025 International Joint Conference on Neural Networks (IJCNN 2025) – VERIMEDIA Workshop (https://sites.unica.it/ff4all/verimediaijcnn/)

The paper introduces RawNetLite, a lightweight deep learning model for detecting audio deepfakes directly from raw waveforms. Unlike traditional methods that rely on handcrafted features, RawNetLite processes raw audio end-to-end, using convolutional and recurrent layers to capture spectral and temporal patterns. To improve generalization, the authors train the model on multiple datasets (FakeOrReal, AVSpoof2021, CodecFake) and apply waveform-level augmentations like pitch shifting and noise. They also use Focal Loss to focus on hard-to-classify examples. Results show excellent in-domain performance (over 99% F1 on FakeOrReal) and strong cross-dataset robustness, especially when using all datasets and augmentations (83.4% F1 on unseen data). The study highlights that diverse training data and smart training strategies are key to building reliable audio deepfake detectors.

Authors: Andrea Di Pierno, Luca Guarnera, Dario Allegra, Sebastiano Battiato

#Audio #deepfake #ff4ll #VERIMEDIA #IJCNN

International Conference in Computer Vision and Pattern Recognition (CVPR25)

Wed, 12 Nov 2025 10:18:38 +0000

Excited to announce that our paper “Exploiting Adversarial Learning and Topology Augmentation for Open-Set Visual Recognition” has been accepted at the International Conference in Computer Vision and Pattern Recognition (CVPR25), within the prestigious Women in Computer Vision (WiCV) Workshop!

Abstract:

This paper proposes a novel approach that introduces a dedicated unknown class, inspired by recent studies on adversarial machine learning images, that are completely unrecognizable to humans yet are confidently misclassified by CNNs as objects. Our method seeks to improve the class cohesion in the feature space, guiding the model to manage the vector space better. We demonstrate that using unknown and unseen images leads to more effective management of the feature space and significantly improves recognition performance in the presence of unknown inputs. Initially, we evaluate the distribution of a C-class subset from the MNIST and ImageNet dataset, investigating the single accuracy of each class, and using t-SNE visualizations to assess how these classes are arranged in the feature space, for a closed-set classification. Then, we extend our approach to a C+1 classification scenario. Indeed, the unknown class is constructed using a customized method that integrates NEAT, a technique for neural network evolution, and defines a novel fitness function for this problem, called fNEAT. Then, the synthetic unknown class is added to the test set of images belonging to unseen classes, within an open-set recognition context. The results are especially relevant for real-world applications, such as open-set biometric authentication in an adversarial environment, an area we intend to explore in our future work.

Authors:

Rosa Zuccarà

Georgia Fargetta

Alessandro Ortis

Sebastiano Battiato

University of Catania – Department of Mathematics and Computer Science

AADD -2025 – Grand Challenge, ACM Multimedia 2025

Wed, 12 Nov 2025 10:17:53 +0000

AADD -2025 – Grand Challenge, ACM Multimedia 2025

We invite you to participate at the first Grand Challenge entitled

Adversarial Attacks on Deepfake Detectors: A Challenge in the Era of AI-Generated Media (AADD-2025)

accepted on ACM Multimedia 2025

https://iplab.dmi.unict.it/mfs/acm-aadd-challenge-2025/

Grand Challenge – Overview

The goal of this challenge is to investigate adversarial vulnerabilities of deepfake detection models by generating adversarial perturbed deepfake images that evade state-of-the-art classifiers while maintaining high visual similarity to the original deepfake content. Given the increasing reliance on deepfake detectors in forensic analysis and content moderation, ensuring their robustness against adversarial attacks is of paramount importance. Over the next 3-5 years, this challenge will contribute to the development of more resilient deepfake detection methodologies, mitigating risks associated with adversarial manipulations.

The aim of this challenge is to expose and address vulnerabilities in current deepfake detection systems by designing adversarial attacks that alter deepfake images—rendering them unrecognizable as synthetic content to 4 proposed classifiers—while preserving high visual similarity to the original images. Participants will be provided with one dataset, divided into sixteen subsets: four GAN-based models and four diffusion-based models in high quality resolution, and four GAN-based models and four diffusion-based models in low quality resolution. Participants must focus on the entire dataset. In addition to developing effective adversarial perturbations, participants are required to submit an abstract that outlines their methodology and attach the adversarial images.

Important Dates

· Grand Challenge Team registration: 15.03.2025 – 15.05.2025

· Test set and classifier release: 10.04.2025

· Grand Challenge solution submission: 30.06.2025

· Grand Challenge Winner Notification: 24.07.2025

· Camera ready – Grand Challenge solutions: 26.08.2025

Organizers: Luca Guarnera, Francesco Guarnera, Orazio Pontorno, Claudio Vittorio Ragaglia, Mirko Casu, Giovanni Puglisi, Zahid Akhtar, Sebastiano Battiato.

DFF-2025 Workshop, ACM Multimedia 2025

Wed, 12 Nov 2025 10:12:49 +0000

DFF-2025 Workshop, ACM Multimedia 2025

We invite you to submit your scientific research to the first Deepfake Forensics: Detection, Attribution, Recognition, and Adversarial Challenges in the Era of AI-Generated MediaWorkshop, accepted on ACM Multimedia 2025 (https://acmmm2025.org/)

https://iplab.dmi.unict.it/mfs/acm-dff-ws-2025/

Workshop – Overview

The rapid advancements in deep learning, particularly in generative models such as Generative Adversarial Networks (GANs) and Diffusion Models (DM), have significantly improved the quality and realism of synthetic media, commonly referred to as deepfakes. While these technologies unlock creative possibilities, they simultaneously raise critical concerns regarding digital content authenticity. Deepfake generation and detection are now at the core of multimedia forensics, requiring robust and generalizable methods to identify manipulated content effectively. This workshop aims to bring together researchers and practitioners from diverse fields, including computer vision, multimedia forensics and adversarial machine learning, to explore emerging challenges and solutions in deepfake detection, attribution, recognition and counter-forensic strategies. Specifically, it will address the limitations of current detection models in generalizing to real-world scenarios, the interpretability of forensic results, and the risks posed by synthetic content. Additionally, the workshop will promote discussions on dataset biases, multimodal deepfake analysis, the forensic ballistics of synthetic media, and the legal and ethical implications of deepfake technology, including regulatory challenges and forensic admissibility in court. Through keynote presentations, research paper presentations, and panel discussions, the workshop will provide a comprehensive overview of state-of-the-art deepfake research while promoting interdisciplinary collaborations to address this pressing societal issue.

Topics of interest in this Special Issue include but are not limited to:

· Deepfake Detection on Images, Video, and Audio

· Multimodal Deepfake Detection

· Deepfake Model Recognition and Attribution

· Forensic Ballistics for Deepfake Analysis

· Adversarial Forensics and Counter-Forensic Techniques

· Generative Models for Deepfake Creation

· Multimodal Datasets for Deepfake Detection and Generation

· Explainability and Interpretability in AI Forensics

· Passive Deepfake Authentication Methods

· Active Deepfake Authentication Methods

· Legal and Ethical Implications of Deepfakes: Detection, Regulation and Accountability

Organizers: Luca Guarnera, Francesco Guarnera, Sebastiano Battiato, Giovanni Puglisi, Zahid Akhtar, Mirko Casu, Orazio Pontorno, Claudio Ragaglia

#acmmultimedia #IPLab #Deepfake

Multimedia Tools and Applications

Wed, 12 Nov 2025 10:11:57 +0000

Glad to inform you that our paper on Cloud Gaming has been published on Multimedia Tools and Applications!

Lorenzo Catania

Filippo Stanco

Sebastiano Battiato

https://doi.org/10.1007/s11042-025-20798-y

IEEE Access

Wed, 12 Nov 2025 10:09:14 +0000

Great news! The paper Cloud Detection Challenge – Methods and Results has been accepted for publication in IEEE Access (Q1, I.F. 3.4)

This work presents the Cloud Detection Challenge, an initiative aimed at improving cloud detection using LiDAR-based ceilometer data. We introduced a novel dataset of backscatter profiles transformed into time-height plots, providing a benchmark for state-of-the-art deep learning models. The challenge attracted teams from around the world, who developed AI-based solutions, including CNNs and Transformers, to push the limits of cloud detection technology.

Authors: Alessio Chisari, Luca Guarnera, Alessandro Ortis, Wladimiro Carlo Patatu, Bruno Casella, Luca Naso, Giuseppe Puglisi, Vincenzo Del Zoppo, Mario Valerio Giuffrida, Sebastiano Battiato

#IPLab #CloudDetection #DeepLearning #AI

IEEE Journal of Biomedical and Health Informatics

Wed, 12 Nov 2025 10:08:03 +0000

Our paper Smoking Detection and Cessation: An Updated Scoping Review of Digital and Mobile Health Technologies was just published in IEEE Journal of Biomedical and Health Informatics (Q1 Computer Science & Health Informatics, IF: 6.7)!

Authors: Mirko Casu, Francesco Guarnera, Giusy Rita Maria La Rosa, prof. Sebastiano Battiato, prof. Pasquale Caponnetto, prof. Riccardo Polosa, and Rosalia Emma.

Smoking remains a global health challenge, and innovation is key. Wearable sensors, apps, and AI show promise, but real-world impact requires better integration and validation. Our review of 49 studies (2019-2023) explored the current landscape and future directions.

This work was a major collaboration across several Università di Catania departments, including Mathematics and Computer Science, Educational Sciences, Medclin Unict and the CoEHAR.

This research was also supported by the Ministry of University and Research (MUR) under the PNRR projects SAMOTHRACE and FAIR.

Read the full open-access paper here: https://doi.org/10.1109/JBHI.2025.3549255

CVPR 2025

Tue, 04 Mar 2025 08:54:55 +0000

We have a paper accepted at CVPR 2025!!!

Gazing Into Missteps: Leveraging Eye-Gaze for Unsupervised Mistake Detection in Egocentric Videos of Skilled Human Activities

Authors: Michele Mazzamuto, Antonino Furnari, Yoichi Sato, Giovanni Maria Farinella

In the CVPR email from Programme Committee:

This year, we received 13,008 valid submissions that underwent the review process. The program committee recommended 2878 papers for acceptance, resulting in an acceptance rate of 22.1%. All but 4 papers were initially evaluated by at least three independent reviewers. In the case of the few papers with only 2 reviews, the AC served as the third reviewer. Following the author rebuttal and reviewer discussions, moderated by an Area Chair (AC), a triplet of three ACs reviewed each paper holistically, considering the reviews, author rebuttal, and reviewer discussions. For challenging cases, additional ACs, Senior ACs, and/or Program Chairs (PCs) were consulted.

Additionally, Area Chairs (ACs) identified a number of highly irresponsible reviewers, those who either abandoned the review process entirely or submitted egregiously low-quality reviews, including some generated by large language models (LLMs). Following a thorough investigation, the Program Chairs (PCs) decided to desk-reject 19 papers authored by confirmed highly irresponsible reviewers, which would have been accepted otherwise, in accordance with the previously communicated CVPR 2025 policies. The affected authors have been informed of this decision.

We would like to express our sincere gratitude to all the reviewers, Area Chairs, and Senior Area Chairs for their invaluable contributions to the review process.

MDPI Journal of Imaging

Tue, 04 Mar 2025 08:53:57 +0000

The paper Deepfake Media Forensics: Status and Future Challenges has been accepted for publication in MDPI Journal of Imaging (Q1, IF 2.7)!

This work is part of the FF4LL project (Detection of Deep Fake Media and Life-Long Media Authentication), where we explore cutting-edge techniques for detecting and authenticating deepfake content.

Our research delves into forensic attribution, deepfake detection in real-world scenarios, and both passive and active authentication methods. In detail, the paper highlights:

The impact of deepfakes on security, ethics, and trust.

Existing forensic techniques for deepfake detection, attribution, and authentication.

Research gaps and challenges in handling real-world deepfakes, including adversarial manipulations and evolving generative techniques.

The importance of passive and active authentication to improve media integrity.

The paper outlines future directions for deepfake forensics, emphasizing continuous learning, forensic explainability, and robust detection strategies to counteract the increasing sophistication of AI-generated manipulations.

Thanks to all the forensic researchers from various Italian institutions who contributed to this study!

Stay tuned for more updates!

Authors: Irene Amerini, Mauro Barni, Sebastiano Battiato, Paolo Bestagini, Giulia Boato, Vittoria Bruni, Roberto Caldelli, Francesco De Natale, Rocco De Nicola, Luca Guarnera, Sara Mandelli, Taiba Majid, Gian Luca Marcialis, Marco Micheletto, Andrea Montibeller, Giulia Orrù, Alessandro Ortis, Pericle Perazzo, Giovanni Puglisi, Nischay Purnekar, Davide Salvi, Stefano Tubaro, Massimo Villari, Domenico Vitulano

#Deepfake #MediaForensics #AI #FF4LL #research #IPLab

NeAIxt Project

Tue, 04 Mar 2025 08:53:21 +0000

The JUCHIPS Horizon project:

Next Generation of edge AI crossing technology field (NEAIXT)

has been approved!!!!

Starting expected date: May 2025

Unict-DMI-IPLAB Budget: 750K euros

Dipartimento di Matematica e Informatica – Unict

Università di Catania

Springer Journal of Social Network Analysis and Mining

Tue, 04 Mar 2025 08:52:43 +0000

We proudly announce that our article “Analyzing Interactions in Donation-Based Live Streaming Platforms: A Multi-Leader-Follower Game Approach” has been accepted for publication in the Springer Journal of Social Network Analysis and Mining (Q1).

Authors: Georgia Fargetta, Alessandro Ortis, Sebastiano Battiato and Laura Rosa Maria Scrimali.

Abstract: This paper investigates the interactions between content creators and viewers in donation-based live streaming platforms. In these social media systems, creators produce their content, while viewers enjoy the live streaming and decide to donate money to creators, making decisions to contribute financial support to the creators. To capture the sequential decision process of the model, we introduce a multi-leader-follower game, in which creators act as the leaders of the game and viewers as the followers. Creators first optimize their performance level and the duration of the streams to maximize their profit. Then, viewers optimize the time spent watching a live stream to maximize their utility. Thus, the first stage of the game models the non-cooperative competition among creators, while the second stage represents the behaviour of viewers deciding on their content demands. We formulate these stages as Nash equilibrium problems and then as variational inequalities. We analyze the existence and uniqueness of the Stackelberg equilibrium. Then, we derive a supervised learning algorithm to estimate the parameters of the Artificial Neural Network model, focused on the advertisement function to properly verify the proposed model in a realistic setting.

Link to preprint paper: https://doi.org/10.21203/rs.3.rs-4533593/v1

Ego4D paper

Tue, 04 Mar 2025 08:52:01 +0000

Ego4D paper has reached more than 1000 citations!

This paper was published at CVPR 2022 – great effort of 15 institutions, including UNICT, to push the frontier of egocentric perception.

Project page: https://ego4d-data.org

IEEE Access

Tue, 04 Mar 2025 08:51:18 +0000

We are very proud to announce that our article “A New Pipeline for Snooping Keystroke Based on Deep Learning Algorithm” has been accepted and already published in the journal IEEE Access (Q1).

Authors: Massimo Orazio Spata, @Valerio Maria Russo, Alessandro Ortis, Sebastiano Battiato

Abstract:

This research focuses on the vulnerability issues related to keystroke logging on a physical computer keyboard, known as Snooping Keystrokes. This category of attacks occurs recording an audio track with a smartphone while typing on the keyboard, and processing the audio to detect individual pressed keys. To address this issue, mathematical wavelet transforms have been tested, and key recognition has been implemented using the inference test of a deep learning model based on a Temporal Convolutional Network (TCN). The novelty of the proposed pipeline lies in its dynamic audio analysis and keystroke recognition, which splits the wave based on audio signal peaks generated by key presses. This approach enables an attack in real-world conditions without knowing the exact number of keystrokes typed by the user. Experimental results for the proposed pipeline show a peak accuracy of 98.3%.

Link to IEEE Access paper: https://ieeexplore.ieee.org/document/10858134…

Source code: email to massimo.spata@unict.it

Special Issue @ Computer Vision and Image Understanding

Tue, 04 Mar 2025 08:50:31 +0000

Call for Papers!

Special Issue @ Computer Vision and Image Understanding

Recent Advances in Assistive Computer Vision and Robotics

More information at:https://www.sciencedirect.com/…/recent-advances-in…

Submission Deadline: 17/02/2025

Guest editors:

Giovanni Maria Farinella, PhD
University of Catania, Catania, Italy
Email: giovanni.farinella@unict.it

Antonino Furnari, PhD
University of Catania, Catania, Italy
Email: antonino.furnari@unict.it

Marco Leo, PhD
National Research Council, Roma, Italy
Email: marco.leo@cnr.it

Francesco Ragusa, PhD
University of Catania, Catania, Italy
Email: francesco.ragusa@unict.it

Gerard Medioni, PhD
University of Southern California, Los Angeles, California, USA
Email: medioni@usc.edu

Mohan Trivedi, PhD
University of California San Diego, La Jolla, California, USA
Email: mtrivedi@eng.uscd.edu

Springer book series Communications in Computer and Information Science

Tue, 04 Mar 2025 08:49:31 +0000

Good News (3)!

The paper entitled “Predictions of Hotel Dynamic Pricing for Revenue Management System” has been accepted for publication in the Springer book series Communications in Computer and Information Science!

First author is Susanna Saitta, PhD Student with a scholarship supported by the local company Triscele.

VISAPP 2025

Tue, 04 Mar 2025 08:48:56 +0000

Good news (2)!

The paper entitled “Benchmarking Neural Rendering Approaches for 3D Reconstruction of Underwater Environments” has been accepted for publication in proceedings of VISAPP 2025

First paper of our master students Salvatore Carota and Alessandro Privitera within a PNRR project of Next Vision!

Other authors: Francesco Ragusa, Antonino Furnari, Daniele Di Mauro, Giovanni Maria Farinella

IET Computer Vision journal

Tue, 04 Mar 2025 08:48:23 +0000

Good news!

The paper entitled “Egocentric Action Anticipation from Untrimmed Videos” has been accepted for publication by the IET Computer Vision journal!

Authors: Ivan Rodin, Antonino Furnari, Giovanni Maria Farinella

VISAPP25

Tue, 04 Mar 2025 08:47:51 +0000

Good news at Christmas time!

Two papers accepted!

“Online Detection of End of Take and Release Actions from Egocentric Videos”

First paper of our master student Alessandro Catinello!

VISAPP25

“CrowdSim++: Unifying Crowd Navigation and Obstacle Avoidance”

First paper of our master student Danilo Leocata – Strong support on the development by Marco J Rosano

ICAART25

Other IPLAB members involved: Antonino Furnari, GMF

End of the year is good – more in 2025!

NeurIPS 2024 in Vancouver

Tue, 04 Mar 2025 08:47:06 +0000

We are at NeurIPS 2024 in Vancouver!

Antonino Furnari and Luigi Seminara are going to present our work on “Differentiable Task Graph Learning: Procedural Activity Representation and Online Mistake Detection from Egocentric Videos” Spotlight!!!

What a blast!

NeurIPS is one of the top conf in our field. First time we are there! A milestone.

Low acceptance rate…very proud of our group!

We are on the map

IEEE International Conference on Image Processing 2024 @ Abu Dhabi

Tue, 04 Mar 2025 08:46:16 +0000

next week IEEE International Conference on Image Processing 2024 @ Abu Dhabi

Invited Speaker with a talk entitled:

Generative AI and Impostor Bias: Innovations, Challenges, and Future Directions in Big Visual Data

– Workshop on ‘Big Visual Data Analytic’

https://lnkd.in/deq-2j4C

Two papers:

On the exploitation of DCT-traces in the generative-AI domain – Orazio Pontorno, Luca Guarnera Sebastiano Battiato

On the cloud detection from backscattered images generated from a lidar-based ceilometer: Current state and opportunities – Alessio Chisari, Alessandro Ortis, Luca Guarnera, Wladimiro Carlo Patatu, Rosaria Ausilia Giandolfo, Emanuele Spampinato, Sebastiano Battiato, Valerio Giuffrida

A special session on ‘Generative AI’ co-rganized with

Zeno Geradts ,Christian Reiss, Alessandro Ortis and Luca Guarnera

Dipartimento di Matematica e Informatica – Unict

MetroXRaine 2024

Thu, 31 Oct 2024 08:48:57 +0000

These days, our group participated in the IEEE MetroXRaine 2024 conference (https://www.metroxraine.org/).
In particular, during the Poster Session in these early days, we presented innovative research such as:

– Efficient Atrophy Mapping: A Single-Step U-Net Approach for Rapid Brain Change Estimation. Authors: Riccardo Raciti, Alessia Rondinella, Lemuel Puglisi, Francesco Guarnera, Daniele Ravi, Sebastiano Battiato;
– SynthBA: Reliable Brain Age Estimation Across Multiple MRI Sequences and Resolutions. Authors: Lemuel Puglisi, Alessia Rondinella, Ermelinda De Meo, Francesco Guarnera, Sebastiano Battiato, Daniele Ravi.
– Acoustic Side Channel Attack for Keystroke Splitting in the Wild. Authors: Massimo Orazio Spata, Valerio Maria Russo, Alessandro Ortis and Sebastiano Battiato.
– A Novel Pipeline Based on Attention Mechanism Within a GNN to Predict New Edges Between Diseases and Genes. Authors: Massimo Orazio Spata, Sabrina Conoci, Alessandro Ortis, Sebastiano Battiato.

and other papers presented in the Oral session:
– Enhancing Crime Scene Investigations Through Virtual Reality and Deep Learning Techniques. Authors: Antonino Zappalà, Luca Guarnera, Vincenzo Rinaldi, Salvatore Livatino, Sebastiano Battiato
– Neural Optimization for Pharmaceutical Transportation Under Stressful Conditions. Authors: Georgia Fargetta, Alessandro Ortis, Sebastiano Battiato

We are also involved as chairs of special sessions accepted at this conference:
– Session 4.2 – Artificial Intelligence, Biometrics and Extended Reality for Criminal Investigation and Digital Forensic. Chairs: Milena Martarelli, Università Politecnica delle Marche, Italy, Luca Guarnera, University of Catania, Italy
– Session 5.1 – Artificial Intelligence and Generative Models for Health. Chairs: Francesco Guarnera, University of Catania, Italy, Daniele Ravi, University of Hertfordshire, University College London, UK.

Finally, tomorrow, during the Poster Session, Alessio Chisari will present the results of the Cloud Detection Challenge (https://www.metroxraine.org/cloud-detection-challenge) organized by Alessio Chisari, Sebastiano Battiato, Luca Guarnera, Alessandro Ortis, Wladimiro Carlo Patatu, Mario Valerio Giuffrida.

Our participation in the IEEE MetroXRaine 2024 is the result of an extensive network of multidisciplinary collaborations, involving researchers and practitioners from various universities and institutions.

NEM Summit 2024

Thu, 31 Oct 2024 08:48:04 +0000

Bruxelles – here we are!

NEM Summit 2024 – “Emerging virtual worlds for a new digital society”

I am going to talk about our research on “Procedure Understanding from Egocentric Videos” tomorrow in the session “AI for Virtual Worlds” – Session sponsored by FAIR (https://fondazione-fair.it)

Programme:

https://nem-initiative.org/nem-summit-2024/

Intelligence Based Medicine Elsevier Journal

Thu, 31 Oct 2024 08:47:13 +0000

“Temporal Convolutional Network on Raman Shift for Human Osteoblast Cells Fingerprint Analysis” has been accepted for publication in the journal Intelligence-Based Medicine. An Elsevier journal with quartile Q2 (best quartile Q1).

Authors: Dario Morganti, Maria Giovanna Rizzo, Massimo Orazio Spata, Salvatore Guglielmino, Barbara Fazio, Sebastiano Battiato and Sabrina Conoci

https://www.sciencedirect.com/…/pii/S2666521224000504

*Dipartimento di Matematica e Informatica University of Catania

*CNR Messina,

*URT LabSens Beyond Nano Messina,

*Dipartimento di Scienze chimiche, Biologiche, Farmaceutiche e Ambientali University of Messina

Abstract

The physiological state and biological characteristics of cells play a crucial role in the study of several biological mechanisms that are at the basis of the life. Raman spectroscopy, a powerful non-destructive technique, has shown promise in providing unique molecular fingerprints of cells based on their vibrational states. However, the high-dimensional and noisy nature of Raman spectra poses significant challenges in precise cell classification. In this study, we present a novel deep learning algorithm tailored for human cells fingerprint assignment through Raman shift analysis. The proposed deep learning framework harnesses the power of Temporal Convolutional Networks (TCN) to efficiently extract and process Raman spectra information. Leveraging a dataset of labeled Raman spectra, the model is trained to learn discriminative features that capture the subtle differences in cell composition and molecular structures in differential states. Additionally, the proposed model enables real-time cell fingerprint prediction, making it highly applicable for high throughput analysis in large-scale experiments. Experimental results demonstrate a peak accuracy of 99%, showcasing the effectiveness and precision of the approach. Overall, the developed deep learning algorithm offers a robust and efficient solution for cell fingerprint assignment through Raman shift analysis, opening new avenues for advancements in physiological and biochemical studies.

ICPR 2024

Mon, 14 Oct 2024 06:18:03 +0000

I’m excited to announce that our paper, “ICPR 2024 Competition on Multiple Sclerosis Lesion Segmentation – Methods and Results”, has been accepted at the International Conference on Pattern Recognition (ICPR 2024)

This report summarizes the outcomes of the MSLesSeg Competition, which challenged participants to develop methods for the automatic segmentation of multiple sclerosis (MS) lesions in MRI scans.

The competition utilized an innovative and annotated dataset, featuring MRI scans of MS patients, including follow-up scans acquired at multiple hospitals. The goal was to develop algorithms capable of autonomously and robustly segmenting lesions across different timepoints and on previously unseen patient cohorts, with no need for manual interaction.

All participants have made innovative contributions, representing a significant step forward in the automatic segmentation of MS lesions

The methods proposed by the teams and the results of the competition will be presented at ICPR 2024, which will take place in Kolkata, India.

The MSLesSeg-dataset will soon be published as open access. Stay tuned for updates on how to access this valuable resource!

The preprint version of the paper is available at: https://arxiv.org/abs/2410.07924

For more information on the challenge: https://iplab.dmi.unict.it/mfs/ms-les-seg/

Authors:

Alessia Rondinella

Francesco Guarnera

Elena Crispino

Giulia Russo

clara di lorenzo

Davide Maimone

Francesco Pappalardo

Sebastiano Battiato

#MSLesSeg #ICPR2024 #AI #DeepLearning #MedicalImaging #MSResearch

Pattern Recognition Letters

Mon, 14 Oct 2024 06:17:26 +0000

We are thrilled to announce the Special Issue titled “Open the Brain: Horizons and Challenges in Brain Image Analysis” on Pattern Recognition Letters!

This Special Issue will focus on critical aspects and innovative solutions in brain-focused medical imaging. Our goal is to demonstrate how AI is overcoming challenges and unlocking new possibilities in diagnosing and treating neurological conditions. It will showcase cutting-edge research, ethical and practical considerations, and provide an overview of AI-driven brain imaging’s current state and future directions.

We invite researchers in AI and brain image analysis to submit their papers.

Topics of Interest Include:

– Machine Learning and Deep Learning Algorithms for Brain Image Analysis

– Pattern Recognition for Early Diagnosis of Neurological Diseases

– Automated Brain Image Segmentation Techniques

– Enhancing Brain Image Quality and Artifact Detection

– Generative Models for Brain Image Synthesis and Augmentation

– Foundation Models in Brain Image Analysis

– Multimodal Data Integration for Brain Pattern Recognition

– Predictive Models for Neurological Disease Progression

– Ethical Considerations and Responsible Development of AI Systems for Brain Imaging

Link to Special Issue: https://www.sciencedirect.com/…/about/call-for-papers…

Important Dates:

Submission Portal Open: May 1st, 2025

Submission Deadline: May 20th, 2025

Acceptance Deadline: September 30th, 2025

Submission :

Papers must be submitted via the Editorial Manager, which will be open for submissions from May 1st, 2025. (https://www.editorialmanager.com/prletters/default2.aspx)

Article Type :

When submitting your manuscript, please select the article type VSI: HCBIA.

Author Guidelines :

https://www.sciencedirect.com/…/publish/guide-for-authors

Guest Editors:

Francesco Guarnera

#BrainImageAnalysis #MedicalImaging #MedicalGenerativeData #NeurologicalDisease #MultimodalData

Health Economics Review

Mon, 14 Oct 2024 06:16:25 +0000

La ‘Computer Science’ ce lo diciamo spesso nei nostri consessi, è pervasiva e può generare valore aggiunto e fornire preziosi strumenti di analisi e di riflessione.

Ne è questo un esempio concreto…cui siamo contenti di aver dato un contributo

One for all? Assessing the quality of Italian hospital care with the “benefit of the doubt”composite indicator methods

Francesco Vidoli, Giacomo Pignataro, Sebastiano Battiato, Francesco Guarnera, Calogero Guccio

Just published on Health Economics Review volume 14, Article number: 83 (2024)

https://healtheconomicsreview.biomedcentral.com/…/s1356…

Università di Catania Politecnico di Milano Universita di Urbino

Dipartimento di Matematica e Informatica – Unict

Computers in Biology and Medicine

Mon, 14 Oct 2024 06:15:35 +0000

I am proud to announce that our work, MITS-GAN: Safeguarding Medical Imaging from Tampering with Generative Adversarial Networks, has been published in Computers in Biology and Medicine (Q1, I.F. 7.0). This work addresses a critical problem in medical imaging: preventing the tampering of CT scans.

In this study, we present MITS-GAN, a novel approach based on generative adversarial networks designed to protect medical images from unauthorized modification. Our method introduces an imperceptible noise into the images, making it difficult for attackers to alter them undetected. We have demonstrated that MITS-GAN outperforms existing solutions by safeguarding image integrity.

This approach ensures the reliability of medical scans, which is crucial in preventing misdiagnosis and medical fraud. We believe this progress will open the way for a safer and more ethical use of AI in healthcare. The models and codes are publicly available for the research community!

Authors: Giovanni Pasqualino, Luca Guarnera, Alessandro Ortis, Sebastiano Battiato

ArXiv preprint: https://arxiv.org/abs/2401.09624

NeurIPS 2024

Mon, 14 Oct 2024 06:14:28 +0000

What? What? YEEE!

Amazing result of our group!

We have a paper accepted at NeurIPS 2024!!!!

Paper Title: Differentiable Task Graph Learning: Procedural Activity Representation and Online Mistake Detection from Egocentric Videos

Authors: Luigi Seminara, Giovanni Maria Farinella, Antonino Furnari

Paper on ArXiv: https://arxiv.org/abs/2406.01486

Code: https://github.com/fpv…/Differentiable-Task-Graph-Learning

Enjoy reading our paper!

PS: There were 15671 valid paper submissions to the NeurIPS Main Track this year, of which the program committee accepted 25.8% for presentation at the conference.

PNRR FAIR

Mon, 14 Oct 2024 06:13:54 +0000

Domani presenteremo a Napoli alcuni dei nostri risultati relativamente al progetto PNRR FAIR!

A parte i 3 lavori pubblicati su CVPR 2024, faremo highlight sui 3 lavori che saranno presentati la prossima settimana alla conferenza ECCV24, come anche alcuni risultati pubblicati su journal quali IJCV/PAMI.

Titolo del mio talk è “Procedure Understanding from Egocentric Videos”. Risultati sono quelli dei nostri straordinari ricercatori del DMI:

https://iplab.dmi.unict.it/fpv/

Felici di esser stati selezionati a presentare nostri risultati!

Vedi programma qui: https://www.fairconference2024.it/index.php/programma

Stanford’s TOP 2% World Sciences

Mon, 14 Oct 2024 06:12:23 +0000

A quanto pare, due IPLabini sono inclusi nella lista https://elsevier.digitalcommonsdata.com/datasets/btchxktzyw/7?fbclid=IwY2xjawF5o6VleHRuA2FlbQIxMAABHcC-aTBAxhyIlerq_ZQbGEc1I7_A2_R76h1mE7Cqz5MBAB0zrxvOHFE6Bg_aem_lPor190uBVMaSeGgVY9nUw

Mon, 14 Oct 2024 06:11:16 +0000

Finnaly out on IEEE Transactions on Pattern Analysis and Machine Intelligence!

(Conference version published at CVPR22 has aready more than 700 citations on Google Scholar – what an impact!)

Stay tuned! News are coming soon about Egocentric Vision Research @ UniCT!https://ieeexplore.ieee.org/document/10611736?fbclid=IwY2xjawF5o2BleHRuA2FlbQIxMQABHW6N8oYi2TsRStMReKGu2cY1ZM3BaLzPZ3o1CkXEUrz0Wp5FzMN2squtEg_aem_qRcQTYXIxpWidDzjoo1IVQ

MICCAI 2024

Mon, 14 Oct 2024 06:10:25 +0000

We are delighted to announce that the Brain Latent Progression (BrLP) paper has been selected for oral presentation at MICCAI 2024!

Authors: Puglisi L., Prof. Daniel C. Alexander, Prof. Daniele Ravì

preprint: https://arxiv.org/abs/2405.03328

code: https://github.com/LemuelPuglisi/BrLP

case studies: https://youtu.be/6YKz2MNM4jg?si=gHrCdH-Aq9hL1RNk

ICPR 2024

Mon, 14 Oct 2024 06:09:48 +0000

Happy to notify that the paper with title “The Analog Layer: Simulating Imperfect Computations in Neural Networks to Improve Robustness and Generalization Ability” has been accepted for publication in proceedings of ICPR 2024!

This work is an extended version of the Bachelor Thesis of our intern student Giovanni Maria Manduca.

Authors: Giovanni Maria Manduca, Antonino Furnari, GMF

ACM SAC 2025

Mon, 14 Oct 2024 06:09:21 +0000

========= ACM SAC 2025 | CALL FOR PAPERS ===

40th ACM/SIGAPP Symposium On Applied Computing (SAC 2025)

March 31 – April 4, 2025

Catania, Italy

https://www.sigapp.org/sac/sac2025/

==========================================

SAC provides a collaborative space to share insights and engage in discussions on the latest developments, trends, experiences, and challenges in applied computing. With a remarkable legacy spanning 39 years, SAC has evolved into a premier international venue bringing together researchers and practitioners addressing real-world problems. SAC 2025 features 42 specialized tracks, each with its own Call for Papers and Program Committee.

=============== CONFERENCE TRACKS ========

** Accessible Devices and Technologies

** Artificial Intelligence for Education

** Artificial Intelligence for Financial Domain

** Bioinformatics

** Cloud Continuum

** Computational Horizons in Medical Imaging

** Cyber-Physical Systems

** Dependable, Adaptive, and Secure Distributed Systems

** Decentralized Applications with Blockchain, DLT, and Crypto-Currencies

** Databases and Big Data Management

** Data Streams

** Data Spaces and Trusted Data Sharing

** Applications of Evolutionary Computation

** Embedded System

** Geographic Information Analytics

** Graph Models for Learning and Recognition

** IoT and Edge Computing

** Innovations in Multimedia Forensics and Biometric Security

** Interoperability

** Intelligent Robotics and Multi-Agent Systems

** Intelligent Systems for Digital Era

** Knowledge and Natural Language Processing

** Knowledge Representation and Reasoning

** Machine Learning and Its Application

** Mobile and Smart Computing

** Computer Networking

** Operating Systems and Software Platforms

** Programming Languages

** Quantum Software Engineering Practices

** Requirements Engineering

** Renewable and Sustainable Energy

** Safe, Secure and Robust Artificial Intelligence

** Software Architecture: Theory, Technology, and Applications

** Smart Cities and Critical Infrastructures

** Software Engineering

** Computer Security

** Social Network and Media Analysis

** Software ProtectIon and Reverse Engineering

** Semantic Technology

** Software Verification and Testing

** Selected Area of Wireless Communications and Networking

** Web Engineering

=============== IMPORTANT DATES ===================

Paper Submission: September 20, 2024

Author Notification: October 30, 2024

Final Paper Submission: November 29, 2024

Author registration: December 6, 2024

=============== TYPES OF SUBMISSION ===============

** Full papers: 8 pages with the option for up to 2 additional pages

** Posters: 2 pages with the option for up to 1 additional page

** Student Research Abstracts: 2 pages

We are looking forward to receiving your contributions!

ACM SAC 2025 Organization Committee

Juw Won Park and Adam Przybylek

Program Co-Chairs

Conference Co-Chair

Jiman Hong, Soongsil University, Seoul, Korea

Conference Co-Chair

Sebastiano Battiato, University of Catania, Catania, Italy

Conference Vice Chair

Christian Esposito, University of Salerno, Italy

ICPR 2024

Mon, 14 Oct 2024 06:08:05 +0000

We are proud to announce that our paper, “DeepFeatureX Net: Deep Features eXtractors based Network for discriminating synthetic from real images“, has just been accepted at the ICPR 2024 conference, which will be held in Kolkata, India, from 1 December to 5 December.

In this work, we address a critical challenge in Digital Forensics: distinguishing between real and AI-generated images, particularly those created by deep learning algorithms known as Deepfakes. Our study introduces an innovative method that improves generalization—the ability to accurately classify images generated by previously unseen architectures.

Our approach involves three specialized blocks, or Base Models, each focusing on extracting discriminative features from images produced by Diffusion Models, GANs, or REAL. By utilizing deliberately unbalanced datasets during training, we enhance the model’s effectiveness. The outputs from these blocks are then combined and processed to determine the image’s origin. The results are promising—our method not only demonstrates strong resilience to JPEG compression and other adversarial attacks but also surpasses current state-of-the-art techniques in generalization tests.

ArXiv preprint: https://arxiv.org/abs/2404.15697

Code, models and dataset are available at https://github.com/opontorno/block-based_deepfake-detection

Authors: Orazio Pontorno, Luca Guarnera, Sebastiano Battiato

#ICPR2024 #Deepfakes #DigitalForensics #AI #MachineLearning #ComputerVision #Research #Kolkata

Intellisys 2024

Mon, 14 Oct 2024 06:07:27 +0000

https://link.springer.com/…/10.1007/978-3-031-66329-1_26

Ci sono alcuni lavori che ti restano nel cuore per un po’, perché ci hai messo tutta la tua passione , il tuo impegno ed un pezzo del tuo tempo per portarli a termine.

La storia di questa pubblicazione inizia il 3 agosto del 2023. Esattamente un anno fa’ quando grazie ad una news diffusa su internet trovo su arXiv questo articolo: A Practical Deep Learning-Based Acoustic Side Channel Attack on Keyboards di Joshua Harrison, Ehsan Toreini, Maryam Mehrnezhad afferenti ad un gruppo di Università Londinesi.

L’articolo dice che registrando un audio con uno smartphone durante la battitura dei tasti di una tastiera ed addestrando un modello di intelligenza artificiale è possibile a partire dall’audio registrato, ricostruire con un accuratezza del 95% quali tasti sono stati cliccati. Dice inoltre che questo sistema funziona anche in video conferenza (zoom) con un accuratezza del 93%.

Decido allora il 3 Agosto dell’anno scorso di scrivere una email ad i Professori Londinesi (a tutti e tre) chiedendo se potevano mandarmi il loro dataset usato per addestrare il modello di AI usato. Il giorno dopo mi risponde il primo autore il Prof Harrison e mi manda il link al dataset augurandomi “Good Luck!” .

Inizia allora una lunga ricerca, conclusasi qualche giorno fa, dove riusciamo a migliorare i risultati ottenuti da Harrison et al, producendo una nuova pipeline ed una demo su smartphone per provare il modello di AI usato.

Sono stati realizzati diversi dataset, per verificare questa vulnerabilità. Inoltre il funzionamento del modello è stato testato su diverse tastiere di laptop e tastiere meccaniche con diversi smartphone (iphone e xiaomi), a diverse posizioni e distanze dei microfoni dello smartphone rispetto alla sorgente audio della tastiera e digitando a diverse velocità e con differente energia. Mesi di test e prove.

I risultati prodotti sullo stesso dataset di Harrison et al mostrano un’accuracy del 98.3% in locale e del 93,05% in video conferenza.

Massimo Orazio Spata, @Valerio Maria Russo, Alessandro Ortis, Sebastiano Battiato

MultimediaForensics Top20

Mon, 14 Oct 2024 06:06:25 +0000

We are proud to announce that we are among the top 20 globally in deepfake research! We will continue to advance our research in the deepfake domain, ensuring ongoing innovation in the multimedia forensic field.

https://publications.eai.eu/…/sis/article/view/4883/3278

#Research #Deepfake #MultimediaForensics #Top20

ICIAP 2025

Mon, 14 Oct 2024 06:05:41 +0000

ICIAP 2025 will be in Rome!

15-19 Sept 2025

https://www.iciap.org/

Paper submission deadline first round: 1 feb 2025

Call for workshops will come soon – Stay tuned!

If you wish to propose a workshop, please do not hesitate to send me an email.

ICPR 2024

Mon, 14 Oct 2024 06:05:14 +0000

We are thrilled to announce the conclusion of the MSLesSeg competition!

We’re excited to share that 28 teams registered for the competition, and 16 of them submitted their results. The final leaderboard is now live on our official challenge website. Check it out here:

https://iplab.dmi.unict.it/mfs/ms-les-seg/

All participants have presented innovative contributions, representing a significant step forward in the automatic segmentation of MS lesions. We are confident that the MSLesSeg dataset will continue to inspire groundbreaking research and the development of advanced algorithmic models to enhance the study of MS lesions.

The competition and the methods proposed by each team will be presented at ICPR 2024. We can’t wait to share the results and innovations that have emerged from this initiative.

Organizers:

Alessia Rondinella Francesco Guarnera Sebastiano Battiato Giulia Russo Francesco Pappalardo Elena Crispino Clara Di Lorenzo Davide Maimone

#MSLesSeg #ICPR2024 #AI #DeepLearning #MedicalImaging #MSResearch

Applied Sciences

Mon, 14 Oct 2024 06:04:16 +0000

We’re excited to announce that our paper “AI Chatbots for Mental Health: A Scoping Review of Effectiveness, Feasibility, and Applications” has been published in the journal Applied Sciences!

In this paper, we present a scoping review of AI chatbots designed to support mental health. Our review explores how these digital tools are being used to address various mental health conditions, from depression and anxiety to substance use disorders and preventive care.

Authors: Mirko Casu, Sergio Triscari, Luca Guarnera, Sebastiano Battiato, Pasquale Caponnetto.

Full paper: https://www.mdpi.com/2076-3417/14/13/5889

ECCV24 workshop

Mon, 14 Oct 2024 06:03:39 +0000

Very happy to be one of the invited speakers of the ECCV24 workshop on “EYES OF THE FUTURE: INTEGRATING COMPUTER VISION IN SMART EYEWEAR” organized by Luxottica and PoliMi!

We have a dream – Egocentric Vision!

https://sites.google.com/…/icvse…/icvse/invited-speakers

ECCV 2024

Mon, 14 Oct 2024 06:03:06 +0000

A W E S O M E !

Three papers accepted at ECCV 2024!

Are Synthetic Data Useful for Egocentric Hand-Object Interaction Detection?

Authors: Rosario Leonardi, Antonino Furnari, Francesco Ragusa, Giovanni Maria Farinella

Pre-print: https://arxiv.org/pdf/2312.02672

____________

Synchronization is All You Need: Exocentric-to-Egocentric Transfer for Temporal Action Segmentation with Unlabeled Synchronized Video Pairs

In collaboration with University of Nottingham

Authors: Camillo Quattrocchi, Antonino Furnari, Daniele Di Mauro, Mario Valerio Giuffrida, Giovanni Maria Farinella

Pre-Print: https://arxiv.org/abs/2312.02638

____________

AFF-ttention! Affordances and Attention models for Short-Term Object Interaction Anticipation

In collaboration with University of Zaragoza

Authors: Lorenzo Mur-Labadia, Ruben Martinez-Cantin, Josechu Guerrero, Giovanni Maria Farinella, Antonino Furnari

Pre-print: https://arxiv.org/pdf/2406.01194

ICVSS 2024 – Numbers

Mon, 14 Oct 2024 06:02:34 +0000

ICVSS – Numbers

First edition in 2007 – ICVSS is 18 years old!

https://iplab.dmi.unict.it/icvss2024/PreviousEditions

7 days to the opening of the 2024 edition

https://iplab.dmi.unict.it/icvss2024/

626 applications from more than 40 countries in 2024

https://iplab.dmi.unict.it/…/ICVSS2024_Applications.png

180 selected attendees via CV and reference letters

15 Keynote Speakers

https://iplab.dmi.unict.it/icvss2024/Committee

1 Reading Group – Students Meeting with Mentors

1 Essay Competition

https://iplab.dmi.unict.it/icvss2024/EssayCompetition

152 Posters in 2 sessions

https://iplab.dmi.unict.it/icvss2024/Posters

1 Industry meets Students Session

https://iplab.dmi.unict.it/icvss2024/IndustrialSession

4 Social Events

https://iplab.dmi.unict.it/icvss2024/Programme

I cannot wait more!

2024 IEEE MetroXRAINE

Mon, 14 Oct 2024 06:01:49 +0000

Good news!

The paper with title “ViLaBot: Connecting Vision and Language for Robots That Assist Humans at Home” has been accepted with minor revisions for publication and presentation at the 2024 IEEE MetroXRAINE!

In this paper we present ViLaBot (Vision and Language roBot), a system designed to aid humans in daily activities while at home. ViLaBot combines a language model with a library of basic visuomotor skills to understand human needs, create action plans and execute them. ViLaBot relies solely on onboard visual and proprioceptive sensing, eliminating the need for pre-built maps or precise object locations and facilitating real-world deployment in a variety of environments.

This is a paper in collaboration with Philips Research Eindhoven within the European Marie Curie project “Philhumans” (https://www.philhumans.eu)

Authors: Asfand Yaar, Marco J Rosano, Aki Harma, Antonino Furnari, Giovanni Maria Farinella

Forensic Science International: Digital Investigation

Mon, 14 Oct 2024 06:01:00 +0000

We are extremely happy to announce that the paper in which we give birth to the “impostor bias” has been published in Forensic Science International: Digital Investigation!

In a world where deepfakes have achieved an exceptionally high level of quality and realism, a new cognitive bias has emerged. This bias stems from the pervasive doubt individuals feel when viewing digital images or videos, or listening to audio. They instinctively question the authenticity of these multimedia elements, wondering if they are real or fake.

We have termed this systematic doubt “impostor bias,“ drawing inspiration from the concept of impostor syndrome.

Title: GenAI Mirage: The Impostor Bias and the Deepfake Detection Challenge in the Era of Artificial Illusions
Authors: Mirko Casu, Luca Guarnera, Sebastiano Battiato, Pasquale Caponnetto.

You can find it here:

https://www.sciencedirect.com/…/pii/S2666281724001197

MICCAI 2024

Mon, 14 Oct 2024 06:00:02 +0000

Happy to announce that our paper “TADM: Temporally-Aware Diffusion Model for Neurodegenerative Progression on Brain MRI” has been accepted at the main conference of MICCAI 2024, the top-tier conference in medical imaging with an acceptance rate of 30% this year.

This work introduces TADM, a novel approach based on diffusion models to accurately infer progression in brain MRIs, by learning the distribution of structural changes in terms of intensity differences between scans at different ages. During training, TADM leverages a pre-trained Brain-Age Estimator (BAE) to guide the diffusion model generation, enhancing its ability to produce accurate MRIs that match the expected age gap between baseline and generated scans.

The preprint arxiv version is available at https://arxiv.org/abs/2406.12411.

Authors: Mattia Litrico (University of Catania), Francesco Guarnera (University of Catania), Valerio Giuffrida (University of Nottingham), Daniele Ravì (University of Hertfordshire), Sebastiano Battiato (University of Catania)

CVPR 2024

Mon, 14 Oct 2024 05:59:16 +0000

What are we going to do @ CVPR 2024, apart attending orals/posters to get new knowledge, enjoy friends, networking with colleagues, exchange ideas, find new collaborations and challenges, enjoy social events and the Seattle local culture?

See our dissemination plan attached – Will be a busy and exciting week!

Antonino Furnari, Francesco Ragusa, Rosario Leonardi, Luigi Seminara

HoloAssist Challenge promoted by Microsoft

Mon, 14 Oct 2024 05:58:48 +0000

Breaking News

We are the WINNERS!

Our Gaze-Based approach for Mistake Detection in egocentric videos has ranked first for the HoloAssist Challenge promoted by Microsoft!

See the pre-print of this new work here: https://arxiv.org/abs/2406.08379

Michele Mazzamuto, Antonino Furnari

Forensic Science International: Digital Investigation

Mon, 14 Oct 2024 05:58:04 +0000

In a world where deepfakes have reached a very high quality and realism, a new cognitive bias has emerged related to the doubt that pervades a person who sees a digital image or video as well as listens to an audio. A priori doubt about the veracity of these multimedia: are they real or fake?

We named this systematic doubt “impostor bias,” from impostor syndrome.

And we are overjoyed to announce that the paper in which we give it birth has been accepted in Forensic Science International: Digital Investigation (Q1 in Law, Medical Lab Technology, Pathology and Forensic Medicine; Q2 in Computer Science Applications and Information Systems)!

Title: GenAI Mirage: The Impostor Bias and the Deepfake Detection Challenge in the Era of Artificial Illusions
Authors: Mirko Casu, Luca Guarnera, Sebastiano Battiato, Pasquale Caponnetto.

While waiting for the final version, you can find v1 on arXiv here:

https://arxiv.org/abs/2312.16220

EgoVis 2024 Workshop

Mon, 14 Oct 2024 05:57:08 +0000

“Affordances and attention based models” score 2nd on the Short-term Anticipation Ego4D Challenge

Presentation will be given at EgoVis workshop @ CVPR 2024 in Seattle on 17 june 2024.

See pre-print of our new AFF-ttention model here:

https://arxiv.org/abs/2406.01194

MetrtoXRaine 2024

Mon, 14 Oct 2024 05:53:11 +0000

We are happy to announce that our paper ‘A Case Study on the Use of Hypnosis Combined With Virtual Reality in Psychological Partner Violence’ has been accepted for publication and presentation at the 2024 IEEE #MetroXRAINE!

Authors: Mirko Casu, Salvatore Bellissima, Giorgia Farruggio, Chiara Farrauto, Pasquale Caponnetto.

MetroXRaine 2024

Mon, 14 Oct 2024 05:52:30 +0000

We are pleased to announce that paper ‘A Novel Pipeline Based on Attention Mechanism Within a GNN to Predict New Edges Between Diseases and Genes’ has been accepted for publication and presentation at the 2024 IEEE MetroXRAINE.

Authors: Massimo Orazio Spata, Sabrina Conoci, Alessandro Ortis, Sebastiano Battiato

MetroXRaine 2024

Mon, 14 Oct 2024 05:51:45 +0000

Good news!

Two papers accepted for publication and presentation at the 2024 IEEE MetroXRAINE

“VALUE: Visual Analysis for Location and Understanding of Environments” in collaboration with Xenia Progetti, Next Vision s.r.l. (https://www.nextvisionlab.it), ICAR-CNR Palermo, Di3A-Cutgana

“A Mixed Reality Application to Help Impaired People Rehabilitate Outside Clinical Environments” in collaboration with A.I.A.S. Acireale (https://www.aiasacireale.it), Next Vision s.r.l., and Maxistore Decò Gruppo Leone

Congrats to Michele Mazzamuto⁩ and co-authors!

ESANN 2024

Mon, 14 Oct 2024 05:50:57 +0000

Another good news this week! Our paper, “Federated Learning in a Semi-Supervised Environment for Earth Observation Data” has been accepted at the European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN 2024), which will be held in Bruges (Belgium) from October 9 to October 11.

In this paper, we propose FedRec, a federated learning workflow taking advantage of unlabelled data in a semi-supervised environment to assist in the training of a supervised aggregated model. In our proposed method, an encoder architecture extracting features from unlabelled data is aggregated with the feature extractor of a classification model via weight averaging. The fully connected layers of the supervised models are also averaged in a federated fashion. We show the effectiveness of our approach by comparing it with the state-of-the-art federated algorithm, an isolated and a centralised baseline, on novel cloud detection datasets (just accepted at ICIP 2024).

Arxiv preprint will be public as soon as possible.

GitHub code repository: https://github.com/CasellaJr/FedRec

Authors: Bruno Casella *, Alessio Barbaro Chisari *, Marco Aldinucci, Sebastiano Battiato, Mario Valerio Giuffrida

*Both authors contributed equally to this work

International Journal of Nanomedicine

Mon, 14 Oct 2024 05:50:21 +0000

Exciting news! Our paper “Nose-to-Brain Drug Delivery and Physico-Chemical Properties of Nanosystems: Analysis and Correlation Studies of Data from Scientific Literature” has been published in the International Journal of Nanomedicine.

Authors: Bonaccorso A, Alessandro Ortis, Musumeci T, Carbone C, Hussain M, Di Salvatore V, Sebastiano Battiato, Francesco Pappalardo, and Pignatello R. #Nanomedicine #DrugDelivery #iplab #unict

EgoVIS 2023/2024

Mon, 14 Oct 2024 05:49:03 +0000

The 10 papers selected as the EgoVis 2022/2023 Distinguished Paper Awards have been announced!

https://egovis.github.io/awards/2022_2023/

The awards ceremony will take place @CVPR EgoVis Workshop 17/06 where authors will give short speeches about the story behind these seminal papers.https://egovis.github.io/awards/2022_2023/?fbclid=IwY2xjawF5nhtleHRuA2FlbQIxMQABHefZ9DWmLR73NCGuZDWCk8SyWxy9BlPawNDJBhW-q7wJTxMXiZTapq7xtg_aem_stceJ-Pj0JF7C_2Hfhy1Nw