We start with a brief introduction of descriptive metadata and its role for discoverability of media content. We then investigate two human-centered AI applications.

First, we will look at facial recognition and discuss how to process, augment, annotate and create bounding boxes around identified public figures/celebrities in images using platform microservices.

Secondly, the audience will be introduced to automatic caption generation using customized sequence-to-sequence model. As an example, we will use sports images from English Premier League.

Finally, we combine both approaches to generate descriptive metadata that replicates human-like style.

About the speaker:

Yulia is working as a Senior Data Scientist at Applied Innovation team at Reuters News agency. For the past year and a half, she’s based in Toronto, Canada. Her work is focused on artificial intelligence applications in news and media industry: in particular, insights extraction from video and image content, continuous speech transcription, and the creation of customized solutions for captioning. Previously, Yulia has been working on big data analytics and computer vision applications in wildlife monitoring. Yulia is Adjunct Professor in Environmental Economics at University of Helsinki, Finland. She has a PhD in Scientific Computing from University of Jyväskylä (Finland) and a PhD Discrete Mathematics and Cybernetics from St. Petersburg University (Russia).

DataTalks.Club is the place to talk about data. Join our slack community!