Mark de Blaauw
A Scalable Logo Recognition Model with Deep Meta-Learning
To help people deal with the vast amount of information they face during attendance of a conference, we want to create an application which can automatically store information from an organisation by taking a picture of its logo. Our goal is to create a scalable logo classifier. To do this, we created a meta-learning classifier that can learn different sets of logo classes with only five training samples per class.
My name is Mark and I recently graduated in Business Analytics at the Vrije universiteit of Amsterdam. At the moment my interests lie in scalable AI-driven applications, such as my thesis subject, which focused on meta-learning. Or a project I am currently working on. Using voice verification to make voice assistants more secure and investigate their vulnerability to adversarial attacks.