Evaluation of Neural Processing Edge Computing Devices for Handwriting Recognition

Background

There are numerous situations in daily life where we take notes and afterwards realize that we need the content in a digital form (like meeting notes, protocols, forms or lecture notes). But how to digitize your handwritten notes? There are different products like document scanners, tablets with a stylus or smart pens. All of these solutions require additional equipment or special paper to work correctly. With this thesis we will explore the

possibilities of improving a system that is able to digitize handwriting with a pen that writes on regular paper. Within this work there will be a strong cooperation with the STABILO International GmbH. A prototype of the DigiPen, a recording system, data sets, developed machine learning methods and evaluation boards will be provided by STABILO.

Tasks

The primary goal of this thesis is the adaptation of an existing neural network on various Neural Processing computing devices for Edge AI devices and systems, such as Google Coral or BrainChip Akida. The different Neural Processing computing devices shall be evaluated with respect to power consumption and compared with a central data processing system. Based on the results, a tradeoff between performance and power consumption should be identified in detail.

Requirements

• Basics of Artificial Intelligence, in particular machine-learning methods
• First experiences with Python
• Motivation and interest in solving technical problems independently