no code implementations • 1 Apr 2024 • Umberto Michieli, Jijoong Moon, Daehyun Kim, Mete Ozay
Nowadays, users demand for increased personalization of vision systems to localize and identify personal instances of objects (e. g., my dog rather than dog) from a few-shot dataset only.
no code implementations • 21 Mar 2024 • Elena Camuffo, Umberto Michieli, Jijoong Moon, Daehyun Kim, Mete Ozay
Improving model robustness in case of corrupted images is among the key challenges to enable robust vision systems on smart devices, such as robotic agents.
no code implementations • 28 Feb 2024 • Hafiz Tiomoko Ali, Umberto Michieli, Ji Joong Moon, Daehyun Kim, Mete Ozay
Inspired by NC properties, we explore in this paper the transferability of DNN models trained with their last layer weight fixed according to ETF.
1 code implementation • 28 Feb 2024 • Francesco Barbato, Umberto Michieli, Mehmet Kerim Yucel, Pietro Zanuttigh, Mete Ozay
To this end, we design a small, modular, and efficient (just 2GFLOPs to process a Full HD image) system to enhance input data for robust downstream multimedia understanding with minimal computational cost.
no code implementations • 28 Feb 2024 • Umberto Michieli, Mete Ozay
Continual Learning (CL) aims to learn a sequence of problems (i. e., tasks and domains) by transferring knowledge acquired on previous problems, whilst avoiding forgetting of past ones.
no code implementations • 19 Sep 2023 • Chang Liu, Giulia Rizzoli, Francesco Barbato, Andrea Maracani, Marco Toldo, Umberto Michieli, Yi Niu, Pietro Zanuttigh
Catastrophic forgetting of previous knowledge is a critical issue in continual learning typically handled through various regularization strategies.
1 code implementation • 24 Jul 2023 • Umberto Michieli, Pablo Peso Parada, Mete Ozay
Keyword Spotting (KWS) models on embedded devices should adapt fast to new user-defined words without forgetting previous ones.
1 code implementation • 24 Jul 2023 • Edward Fish, Umberto Michieli, Mete Ozay
Recent advancement in Automatic Speech Recognition (ASR) has produced large AI models, which become impractical for deployment in mobile devices.
Automatic Speech Recognition Automatic Speech Recognition (ASR) +2
no code implementations • 19 Jul 2023 • Umberto Michieli, Mete Ozay
Vision systems mounted on home robots need to interact with unseen classes in changing environments.
no code implementations • 26 Jan 2023 • Elena Camuffo, Umberto Michieli, Simone Milani
Recent advances in autonomous robotic technologies have highlighted the growing need for precise environmental analysis.
no code implementations • 13 Oct 2022 • Marco Toldo, Umberto Michieli, Pietro Zanuttigh
Then, we address the proposed setup by using style transfer techniques to extend knowledge across domains when learning incremental tasks and a robust distillation framework to effectively recollect task knowledge under incremental domain shift.
1 code implementation • 5 Oct 2022 • Donald Shenaj, Eros Fanì, Marco Toldo, Debora Caldarola, Antonio Tavera, Umberto Michieli, Marco Ciccone, Pietro Zanuttigh, Barbara Caputo
Federated Learning (FL) has recently emerged as a possible way to tackle the domain shift in real-world Semantic Segmentation (SS) without compromising the private nature of the collected data.
no code implementations • 20 Apr 2022 • Paolo Testolina, Francesco Barbato, Umberto Michieli, Marco Giordani, Pietro Zanuttigh, Michele Zorzi
Accurate scene understanding from multiple sensors mounted on cars is a key requirement for autonomous driving systems.
no code implementations • 18 Jan 2022 • Donald Shenaj, Francesco Barbato, Umberto Michieli, Pietro Zanuttigh
In this paper we introduce the novel task of coarse-to-fine learning of semantic segmentation architectures in presence of domain shift.
1 code implementation • ICCV 2021 • Andrea Maracani, Umberto Michieli, Marco Toldo, Pietro Zanuttigh
Replay data are then blended with new samples during the incremental steps.
1 code implementation • 6 Aug 2021 • Francesco Barbato, Umberto Michieli, Marco Toldo, Pietro Zanuttigh
Deep learning models obtain impressive accuracy in road scenes understanding, however they need a large quantity of labeled samples for their training.
no code implementations • 19 May 2021 • Umberto Michieli, Mete Ozay
Federated Learning (FL) is a framework which enables distributed model training using a large corpus of decentralized training data.
1 code implementation • 6 Apr 2021 • Francesco Barbato, Marco Toldo, Umberto Michieli, Pietro Zanuttigh
Deep convolutional neural networks for semantic segmentation achieve outstanding accuracy, however they also have a couple of major drawbacks: first, they do not generalize well to distributions slightly different from the one of the training data; second, they require a huge amount of labeled data for their optimization.
no code implementations • CVPR 2021 • Umberto Michieli, Pietro Zanuttigh
Second, features sparsification allows to make room in the latent space to accommodate novel classes.
Ranked #3 on Disjoint 15-5 on PASCAL VOC 2012
1 code implementation • 25 Nov 2020 • Marco Toldo, Umberto Michieli, Pietro Zanuttigh
Deep learning frameworks allowed for a remarkable advancement in semantic segmentation, but the data hungry nature of convolutional networks has rapidly raised the demand for adaptation techniques able to transfer learned knowledge from label-abundant domains to unlabeled ones.
no code implementations • ECCV 2020 • Umberto Michieli, Edoardo Borsato, Luca Rossi, Pietro Zanuttigh
To tackle part-level ambiguity and localization we propose a novel adjacency graph-based module that aims at matching the relative spatial relationships between ground truth and predicted parts.
Ranked #7 on Semantic Segmentation on FMB Dataset
no code implementations • 21 May 2020 • Marco Toldo, Andrea Maracani, Umberto Michieli, Pietro Zanuttigh
The aim of this paper is to give an overview of the recent advancements in the Unsupervised Domain Adaptation (UDA) of deep networks for semantic segmentation.
no code implementations • 27 Apr 2020 • Teo Spadotto, Marco Toldo, Umberto Michieli, Pietro Zanuttigh
We introduce a novel UDA framework where a standard supervised loss on labeled synthetic data is supported by an adversarial module and a self-training strategy aiming at aligning the two domain distributions.
no code implementations • 14 Jan 2020 • Marco Toldo, Umberto Michieli, Gianluca Agresti, Pietro Zanuttigh
The supervised training of deep networks for semantic segmentation requires a huge amount of labeled real world data.
no code implementations • 8 Nov 2019 • Umberto Michieli, Pietro Zanuttigh
Deep learning architectures have shown remarkable results in scene understanding problems, however they exhibit a critical drop of performances when they are required to learn incrementally new tasks without forgetting old ones.
no code implementations • 2 Sep 2019 • Umberto Michieli, Matteo Biasetton, Gianluca Agresti, Pietro Zanuttigh
A recently proposed workaround is to train deep networks using synthetic data, but the domain shift between real world and synthetic representations limits the performance.
2 code implementations • 31 Jul 2019 • Umberto Michieli, Pietro Zanuttigh
To tackle this task we propose to distill the knowledge of the previous model to retain the information about previously learned classes, whilst updating the current model to learn the new ones.
Ranked #4 on Domain 11-1 on Cityscapes
no code implementations • 22 Apr 2018 • Umberto Michieli
Who are the most significant players in the history of men tennis?