whoami

Hey there! I'm Francesco Pelosin, I'm Italian and I'm from Bassano del Grappa (north-east). I got a Ph.D. (july 2022) from Ca' Foscari University of Venice where I focused my studies on Continual Learning for Computer Vision. I'm passionate about my work and consider myself an "advanced script k1ddie" :D. If you need to contact me, drop an e-mail here: pelosinf[at]gmail

My path of studies/work 🎓

projects

publications

• 

  Bounding Box-Guided Diffusion for Synthesizing Industrial Images and Segmentation Maps

  Alessandro Simoni*, Francesco Pelosin*

  Synthetic Data for Computer Vision Workshop (CVPR 2025)

  💻 code | 📄 pdf | 📌 cite | 💭 abstract

  @inproceedings{simoni2025syth,
              title={Bounding Box-Guided Diffusion for Synthesizing Industrial Images and Segmentation Maps},
              author={Simoni, Alessandro and Pelosin, Francesco},
              booktitle={Computer Vision and Pattern Recognition Workshops 2025 (CVPR 2025)},
              year={2025}
            }

  Synthetic dataset generation in Computer Vision, particularly for industrial applications, is still underexplored. Industrial defect segmentation, for instance, requires highly accurate labels, yet acquiring such data is costly and time-consuming. To address this challenge, we propose a novel diffusion-based pipeline for generating high-fidelity industrial datasets with minimal supervision. Our approach conditions the diffusion model on enriched bounding box representations to produce precise segmentation masks, ensuring realistic and accurately localized defect synthesis. Compared to existing layout-conditioned generative methods, our approach improves defect consistency and spatial accuracy. We introduce two quantitative metrics to evaluate the effectiveness of our method and assess its impact on a downstream segmentation task trained on real and synthetic data. Our results demonstrate that diffusion-based synthesis can bridge the gap between artificial and real-world industrial data, fostering more reliable and cost-efficient segmentation models. The code is publicly available at https://github.com/covisionlab/diffusion_labeling/
• 

  Dynamic Label Injection for Imbalanced Industrial Defect Segmentation

  Emanuele Caruso*, Francesco Pelosin*, Alessandro Simoni*, Marco Boschetti

  VISION Workshop (ECCV 2024)

  💻 code | 📄 pdf | 📌 cite | 💭 abstract

  @inproceedings{caruso2024dynamic,
              title={Dynamic Label Injection for Imbalanced Industrial Defect Segmentation},
              author={Caruso, Emanuele and Pelosin, Francesco and Simoni, Alessandro and Boschetti, Marco},
              booktitle={Proceedings of the European Conference on Computer Vision Workshops},
              year={2024}
            }

• 

  DUCK: Distance-based Unlearning via Centroid Kinematics

  Marco Cotogni*, Jacopo Bonato*, Luigi Sabetta*, Francesco Pelosin, Alessandro Nicolosi

  Under Revision

  💻 code | 📄 pdf | 📌 cite | 💭 abstract

  @misc{cotogni2023duck,
              title={DUCK: Distance-based Unlearning via Centroid Kinematics}, 
              author={Marco Cotogni and Jacopo Bonato and Luigi Sabetta and Francesco Pelosin and Alessandro Nicolosi},
              year={2023},
              eprint={2312.02052},
              archivePrefix={arXiv},
              primaryClass={cs.CV}
              }

• 

  MIND: Multi-Task Incremental Network Distillation

  Jacopo Bonato*, Francesco Pelosin*, Luigi Sabetta*, Alessandro Nicolosi

  AAAI (2024)

  💻 code | 📄 pdf | 📌 cite | 💭 abstract

  @misc{mind2024,
              author={Bonato, Jacopo and Pelosin, Francesco and Sabetta, Luigi and Nicolosi, Alessandro},
              title = {MIND: Multi-Task Incremental Network Distillation},
              publisher = AAAI Conference on Artificial Intelligence (AAAI 2024),
              year = {2024},
              }

• 

  Simpler Is Better: off-the-shelf Continual Learning Through Pretrained Backbones

  Francesco Pelosin

  Workshop T4V (CVPR 2022)

  💻 code | 📄 pdf | 📌 cite | 💭 abstract

  @misc{pelosin2022simpler,
              author = {Pelosin, Francesco},
              title = {Simpler is Better: off-the-shelf Continual Learning Through Pretrained Backbones},
              publisher = Transformers 4 Vision Workshop (CVPR 2022),
              year = {2022},
              }

• 

  Towards Exemplar-Free Continual Learning in Vision Transformers: an Account of Attention, Functional and Weight Regularization

  Francesco Pelosin*, Saurav Jha*, Andrea Torsello, Bogdan Raducanu, Joost van de Weijer

  Workshop CLVISION (CVPR 2022) - Runner-Up Best Paper Award

  💻 code | 📄 pdf | 📌 cite | 💭 abstract

  @inproceedings{pelosin2022towards,
              title={Towards exemplar-free continual learning in vision transformers: an account of attention, functional and weight regularization},
              author={Pelosin, Francesco and Jha, Saurav and Torsello, Andrea and Raducanu, Bogdan and van de Weijer, Joost},
              booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
              booksubtitle={Workshop on Continual Learning},
              pages={3820--3829},
              year={2022}
          }

• 

  Smaller is Better: An Analysis of Instance Quantity/Quality Trade-off in Rehearsal-based Continual Learning

  Francesco Pelosin, Andrea Torsello

  IJCNN (2022)

  💻 code | 📄 pdf | 📌 cite | 💭 abstract

  @misc{pelosin2021better,
              title={Smaller Is Better: An Analysis of Instance Quantity/Quality Trade-off in Rehearsal-based Continual Learning}, 
              author={Francesco Pelosin and Andrea Torsello},
              year={2022},
              book={IJCNN}
          }
          

• 

  Unsupervised semantic discovery through visual patterns detection

  Francesco Pelosin, Andrea Gasparetto, Andrea Albarelli, Andrea Torsello

  S+SSPR (2022)

  💻 code | 📄 pdf | 📌 cite | 💭 abstract

  @article{DBLP:journals/corr/abs-2102-12213,
                  author    = {Francesco Pelosin and
                                  Andrea Gasparetto and
                                  Andrea Albarelli and
                                  Andrea Torsello},
                  title     = {Unsupervised semantic discovery through visual patterns detection},
                  journal   = {S+SSPR},
                  year      = {2021},
                  }
          

• 

  Separating Structure from Noise in Large Graphs Using the Regularity Lemma

  Marco Fiorucci, Francesco Pelosin, Marcello Pelillo

  Pattern Recognition (2020)

  💻 code | 📄 pdf | 📌 cite | 💭 abstract

  @article{FIORUCCI2020107070,
              title = {Separating Structure from Noise in Large Graphs Using the Regularity Lemma},
              journal = {Pattern Recognition},
              volume = {98},
              pages = {107070},
              year = {2020},
              issn = {0031-3203},
              doi = {https://doi.org/10.1016/j.patcog.2019.107070},
              url = {https://www.sciencedirect.com/science/article/pii/S0031320319303711},
              author = {Marco Fiorucci and Francesco Pelosin and Marcello Pelillo},
              keywords = {Regularity lemma, Graph summarization, Structural patterns, Noise, Randomness, Graph similarity search}
              }
          
              

  How can we separate structural information from noise in large graphs? To address this fundamental question, we propose a graph summarization approach based on Szemerédi's Regularity Lemma, a well-known result in graph theory, which roughly states that every graph can be approximated by the union of a small number of random-like bipartite graphs called "regular pairs". Hence, the Regularity Lemma provides us with a principled way to describe the essential structure of large graphs using a small amount of data. Our paper has several contributions: (i) We present our summarization algorithm which is able to reveal the main structural patterns in large graphs. (ii) We discuss how to use our summarization framework to efficiently retrieve from a database the top-k graphs that are most similar to a query graph. (iii) Finally, we evaluate the noise robustness of our approach in terms of the reconstruction error and the usefulness of the summaries in addressing the graph search task.
• 

  Graph Compression Using The Regularity Method

  Francesco Pelosin

  arxiv (2019) M.Sc. Thesis

  💻 code | 📄 pdf | 📌 cite | 💭 abstract

  @article{DBLP:journals/corr/abs-1810-07275,
                  author    = {Francesco Pelosin},
                  title     = {Graph Compression Using The Regularity Method},
                  journal   = {CoRR},
                  volume    = {abs/1810.07275},
                  year      = {2018},
                  url       = {http://arxiv.org/abs/1810.07275},
                  eprinttype = {arXiv},
                  eprint    = {1810.07275},
                  timestamp = {Tue, 30 Oct 2018 20:39:56 +0100},
                  biburl    = {https://dblp.org/rec/journals/corr/abs-1810-07275.bib},
                  bibsource = {dblp computer science bibliography, https://dblp.org}
                  }
                        
              
          

  I will provide the abstract in a while :D

      *           .          
      .    .                        .
  .             *             *.          
  *     *                 *                         
  .