Neural Modelling of Event Camera Streams for Biosignal Analysis

Masterarbeit

Event cameras, sometimes called neuromorphic cameras, are an alternative to regular frame-based cameras and have been applied in autonomic driving and drone control. In contrary, to frame-based cameras, they have a high dynamic range, can resolve small time intervals, and provide a sparse data representation. So far, their usefulness in medical applications has not been tested extensively.

Tasks:

• Estimate biosignals from event camera stream

• Continue measuring subjects based on existing measurement plan to create a dataset

• Develop a data representation suitable for machine learning

• Adapt existing deep learning architectures to the problem

• Data preparation and documentation for further research