Neuromorphic visual sensing and computing

About the Project

Neuromorphic vision sensors enable the acquisition of scenes with limited energy requirements and low data rate, only acquiring information on what changes in the scene, with a behaviour similar to the one of the human eye. The acquired information can be easily exchanged in internet of things scenarios where battery operated visual sensors are connected to monitor a scene or perform a task requiring limited delay (e.g., tele operated driving).

The project will study lossy compression methods for the data acquired with such methodologies, as well as subjective and task-based criteria for quality assessment of the compressed information. The latter task may leverage machine learning strategies, hence a background in the areas of machine learning is desirable. Also, edge-computing supported compression and processing strategies will be addressed.

The possibility to acquire information from multiple neuromorphic vision sensors will also be considered, as well as application in real life scenarios (e.g., smart cities, healthcare, gaming).

Collaboration with national and international partners (research bodies/academy/industry) is expected, as well as the possibility of an industrial internship during the PhD.

The project will benefit from the state-of-the-art equipment available in our Centre for Augmented and Virtual Environments (CAVE lab) including neuromorphic vision sensors, GPU clusters, light field displays, VR/XR devices, eye trackers.

The selected candidate will become part of the Wireless Multimedia Networking research group, one of the largest and most successful in the Faculty.

Candidates are expected to have at least a BSc 2:1 in a relevant area (computer science, engineering, statistics, mathematics) and possibly an MSc in a related area.

Funding Notes

there is no funding for this project