Francesco Barbato | Ph.D. Student

Phone (office): +39 049 827 7774


Francesco Barbato is a 2nd year Ph.D. Student in Information Engineering at the Department of Information Engineering of the University of Padua.
He received his Bachelor’s degree in Information Engineering in 2018 and his Master’s Degree (with honors) in ICT for Internet and Multimedia in 2020, both from the University of Padova. Afterward, he was awarded a research grant and spent a year in the LTTM laboratory as a collaborator before joining it as a Ph.D. student in 2021.
His research interests and activities regard Computer Vision and Deep Learning for scene understanding (Domain Adaptation, Continual Learning, and Multimodal Learning) with a particular focus on autonomous driving scenarios.

Research areas

Semantic Segmentation

Image Understanding task dealing with the dense labeling of input data, during my research I investigated both 2D (images), 3D (point clouds), and Multi-Modal settings.

Continual Learning

This setup implies the training in subsequent steps of a neural network, each time adding new classes to be recognized. I have investigated this technique in Semantic Segmentation.

Unsupervised Domain Adaptation

This task deals with the adaptation of a neural network to a new domain without requiring labeled samples from it. It has been widely investigated in Semantic Segmentation due to the prohibitive labeling costs.

Multi-Modal Learning

This setting investigates the possibility of using neural network with heterogeneous input data. Common examples are RGB-Depth or RGB-Thermal, recently there has been a lot of interest in the 2D-3D case.



Testolina, Paolo; Barbato, Francesco; Michieli, Umberto; Giordani, Marco; Zanuttigh, Pietro; Zorzi, Michele

SELMA: SEmantic Large-Scale Multimodal Acquisitions in Variable Weather, Daytime and Viewpoints Journal Article

In: IEEE Transactions on Intelligent Transportation Systems, pp. 1–13, 2023.

Links | BibTeX

Barbato, Francesco; Rizzoli, Giulia; Zanuttigh, Pietro

DepthFormer: Multimodal Positional Encodings and Cross-Input Attention for Transformer-Based Segmentation Networks Proceedings Article

In: ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2023.


Rizzoli, Giulia; Barbato, Francesco; Caligiuri, Matteo; Zanuttigh, Pietro

SynDrone-Multi-Modal UAV Dataset for Urban Scenarios Proceedings Article

In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 2210–2220, 2023.


Barbato, Francesco; Camuffo, Elena; Milani, Simone; Zanuttigh, Pietro

Continual Road-Scene Semantic Segmentation via Feature-Aligned Symmetric Multi-Modal Network Journal Article

In: arXiv preprint arXiv:2308.04702, 2023.

Abstract | Links | BibTeX

Liu, Chang; Rizzoli, Giulia; Barbato, Francesco; Michieli, Umberto; Niu, Yi; Zanuttigh, Pietro

RECALL+: Adversarial Web-based Replay for Continual Learning in Semantic Segmentation Journal Article

In: arXiv preprint arXiv:2309.10479, 2023.



Rizzoli, Giulia; Barbato, Francesco; Zanuttigh, Pietro

Multimodal Semantic Segmentation in Autonomous Driving: A Review of Current Approaches and Future Perspectives Journal Article

In: Technologies, vol. 10, no. 4, pp. 90, 2022.

Links | BibTeX

Shenaj, Donald; Barbato, Francesco; Michieli, Umberto; Zanuttigh, Pietro

Continual coarse-to-fine domain adaptation in semantic segmentation Journal Article

In: Image and Vision Computing, vol. 121, pp. 104426, 2022.

Links | BibTeX


Barbato, Francesco; Michieli, Umberto; Toldo, Marco; Zanuttigh, Pietro

Adapting Segmentation Networks to New Domains by Disentangling Latent Representations. Journal Article

In: CoRR, 2021.


Barbato, Francesco; Toldo, Marco; Michieli, Umberto; Zanuttigh, Pietro

Latent space regularization for unsupervised domain adaptation in semantic segmentation Proceedings Article

In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 2835–2845, 2021.


Barbato, Francesco; Michieli, Umberto; Toldo, Marco; Zanuttigh, Pietro

Road scenes segmentation across different domains by disentangling latent representations Journal Article

In: arXiv preprint arXiv:2108.03021, 2021.