This work introduces a next-generation smart city platform using a novel embedded vision system. i.e., HAura. The HAura system integrates a dual camera and other sensors with a powerful embedded computing unit. The powerful perception stack, based on robust deep learning and computer vision techniques, provides a perfect baseline for implementing a variety of security, traffic management and urban planning policies. Choosing to process images directly on the device and transmit only metadata ensures compliance with privacy and security, as well as bandwidth efficiency. The next evolution of the vision stack will finally improve capabilities by introducing a new multi-task perception model.
On-the-Edge Inference Enabled Vision System for Smart Cities / Scribano, Carmelo; Sanudo Olmedo, Ignacio; Verucchi, Micaela; Pani Paudel, Danda; Bertogna, Marko; Van Gool, Luc. - (2025), pp. 24-48. ( SMART 2025, The Fourteenth International Conference on Smart Cities, Systems, Devices and Technologies Valencia, Spain 06-10/04/2024).
On-the-Edge Inference Enabled Vision System for Smart Cities
Carmelo Scribano
;Ignacio Sanudo Olmedo;Micaela Verucchi;Marko Bertogna;
2025
Abstract
This work introduces a next-generation smart city platform using a novel embedded vision system. i.e., HAura. The HAura system integrates a dual camera and other sensors with a powerful embedded computing unit. The powerful perception stack, based on robust deep learning and computer vision techniques, provides a perfect baseline for implementing a variety of security, traffic management and urban planning policies. Choosing to process images directly on the device and transmit only metadata ensures compliance with privacy and security, as well as bandwidth efficiency. The next evolution of the vision stack will finally improve capabilities by introducing a new multi-task perception model.| File | Dimensione | Formato | |
|---|---|---|---|
|
smart_2025_1_50_40031.pdf
Open access
Tipologia:
VOR - Versione pubblicata dall'editore
Dimensione
1.83 MB
Formato
Adobe PDF
|
1.83 MB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate

I metadati presenti in IRIS UNIMORE sono rilasciati con licenza Creative Commons CC0 1.0 Universal, mentre i file delle pubblicazioni sono rilasciati con licenza Attribuzione 4.0 Internazionale (CC BY 4.0), salvo diversa indicazione.
In caso di violazione di copyright, contattare Supporto Iris




