Developing Low-Power Spiking Neural Networks for Image Enhancement on Embedded Neuromorphic Boards: A Multimodal Sensor Fusion Approach

Image enhancement consists of creating a composite image (fused image) from different sources. The objective of this project consists of acquiring images from Electro-Optical sensors and Infrared (IR) sensors in low light conditions, to enhance them, to combine (fuse) them in order to create an image of better quality. To do so, the project will focus on establishing the fundamentals steps to develop Spiking Neural Networks (resources, tests and activities to develop a complete prototype), with very low consumption for embedded neuromorphic boards for an image-enhancement use case.

Faculty Supervisor:

Sean Wood;Jean Rouat;Fabien Alibart

Student:

Partner:

Thales Recherche et Technologie

Discipline:

Computer science

Sector:

Management of companies and enterprises; Manufacturing; Professional, scientific and technical services

University:

Université de Sherbrooke

Program: