Embedded Files
Guido Borghi
He is an Associate Professor at the University of Modena and Reggio Emilia. His research topics mainly belong to Computer Vision and Artificial Intelligence. In particular, he works in the field of Face Analysis (Face Recognition, Head Pose Estimation), Natural User Interfaces (Gesture and Activity Recognition), applied to Biometrics (Morphing Attack Detection), Robotics (Robot Pose Estimation) and Automotive (Driver Monitoring). He is Associate Editor of Simulation & Gaming and Guest Editor of CVIU. He co-organized several workshops (T-CAP@ICIAP, ICPR, ECCV, and WCPA@ECCV) and a tutorial at IJCB 2023 in the field of human and face analysis. He regularly serves as a reviewer for international conferences and journals, such as TPAMI, TIST, TBIOM, CVPR, ECCV and ICCV.
TALK TITLE: Efficient Deep Learning for Embedded Vision: Lightweight Models and Training Strategies
ABSTRACT: Recent advances in computer vision have been remarkable, driven by increasingly sophisticated deep learning architectures and the availability of massive datasets. These breakthroughs, however, come at the cost of intensive data collection and the need for powerful hardware, all requirements that are often incompatible with embedded systems. Indeed, in embedded devices, memory is limited, and computational resources are constrained due to the use of low-power chips, making on-device training impractical and even inference a significant challenge. This talk explores strategies to bring advanced computer vision capabilities to embedded platforms by leveraging lightweight and efficient neural network architectures, including binarized and quantized models. We will also discuss alternative training paradigms, such as continual learning, that enable models to incrementally adapt to new data while respecting tight resource budgets.
Contacts
Rita Delussu <rdelussu@uniss.it>
Naser Damer <naser.damer@igd.fraunhofer.de>
Giorgio Fumera <fumera@unica.it>
Emanuele Ledda <emanuele.ledda@unica.it>
Lorenzo Putzu <lorenzo.putzu@unica.it>
Fabio Roli <fabio.roli@unige.it>
Page updated
Google Sites
Report abuse