Publications

You can also find my articles on my Google Scholar profile.

2026

MARS: a multimodal alignment and ranking system for few-shot segmentation

Published in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2026

A novel plug and play rankiing system to be run on top of existing few-shot segmentation methods to boos their performance thanks to novel multimodal similarity measures.

Recommended citation: Catalano, N., Samele, S., Pertino, P., Matteucci, M. (2026). "Scaling anomaly detection with segmentation models." Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2026.
Paper Link | Download Bibtex | Code

2025

Scaling Anomaly Detection with Segmentation Models

Published in Digital Signal Processing, 2025

This paper investigates the adoption of MaskRCNN segmentation model to scale the task of Anomaly Detection over industrial product images to higher resolutions

Recommended citation: Samele, S., Attore, F., Matteucci, M. (2025). "Scaling anomaly detection with segmentation models." 2025 Digital Signal Processing 105613.
Paper Link | Download Bibtex | Code

2024

Self-Supervised Vision Transformers for Scalable Anomaly Detection over Images

Published in 2024 International Joint Conference on Neural Networks (IJCNN), 2024

This paper explores the potential of scaling anomaly detection to high resolution images using self-supervised Vision Transformers

Recommended citation: Samele, S., Matteucci, M. (2024). "Self-Supervised Vision Transformers for Scalable Anomaly Detection over Images." 2024 International Joint Conference on Neural Networks (IJCNN). pp. 1–10.
Paper Link | Download Bibtex

2022

SGDE: Secure generative data exchange for cross-silo federated learning

Published in 5th International Conference on Artificial Intelligence and Pattern Recognition, 2022

This work proposes SGDE, a privacy-preserving data exchange technique for federated learning across silos.

Recommended citation: Lomurno, E., Archetti, A., Cazzella, L., Samele, S., Di Perna, L., Matteucci, M. (2022). "SGDE: Secure generative data exchange for cross-silo federated learning." Proceedings of the 5th International Conference on Artificial Intelligence and Pattern Recognition. pp. 205–214.
Paper Link | Download Bibtex | Code

Patchwise Sparse Dictionary Learning from pre-trained Neural Network Activation Maps for Anomaly Detection in Images

Published in 2022 26th International Conference on Pattern Recognition (ICPR), 2022

This method detects image anomalies using sparse dictionary learning on activation maps of pre-trained neural networks.

Recommended citation: Samele, S., Matteucci, M. (2022). "Patchwise Sparse Dictionary Learning from pre-trained Neural Network Activation Maps for Anomaly Detection in Images." 2022 26th International Conference on Pattern Recognition (ICPR). pp. 1307–1313.
Paper Link | Download Bibtex | Code

Popnasv2: An efficient multi-objective neural architecture search technique

Published in 2022 International Joint Conference on Neural Networks (IJCNN), 2022

PopNASv2 improves neural architecture search with a multi-objective perspective focusing on efficiency and performance.

Recommended citation: Falanti, A., Lomurno, E., Samele, S., Ardagna, D., Matteucci, M. (2022). "Popnasv2: An efficient multi-objective neural architecture search technique." 2022 International Joint Conference on Neural Networks (IJCNN). pp. 1–8.
Paper Link | Download Bibtex | Code

Trust-no-pixel: A remarkably simple defense against adversarial attacks based on massive inpainting

Published in 2022 International Joint Conference on Neural Networks (IJCNN), 2022

We introduce a defense method against adversarial attacks using massive image inpainting, showing strong robustness.

Recommended citation: Ughini, G., Samele, S., Matteucci, M. (2022). "Trust-no-pixel: A remarkably simple defense against adversarial attacks based on massive inpainting." 2022 International Joint Conference on Neural Networks (IJCNN). pp. 1–10.
Paper Link | Download Bibtex | Code

2021

Pareto-optimal progressive neural architecture search

Published in Proceedings of the Genetic and Evolutionary Computation Conference Companion, 2021

This paper presents a Pareto-optimal approach to progressive neural architecture search, balancing multiple objectives.

Recommended citation: Lomurno, E., Samele, S., Matteucci, M., Ardagna, D. (2021). "Pareto-optimal progressive neural architecture search." Proceedings of the Genetic and Evolutionary Computation Conference Companion. pp. 1726–1734.
Paper Link | Download Bibtex

Stefano Samele