22.01.2024

 

The MCML-Exhibition at Deutsches Museum Munich

Our exhibition was a great success - A short recap

Around 1.000 visitors per day passed by the Science Communication Lab at Deutsches Museum and saw our exhibition during the last three weeks. It was a lot of fun and we are happy that so many visitors were interested and got a glimpse of our research.

One of the exhibits called AMELIE taught visitors the basics of machine learning in a fun and interactive way. The exhibit was built by the LMU Media Informatics Group of our PI Albrecht Schmidt. You can see a video of the exhibit on YouTube.

In addition we had a demo showing how it is possible to simultaneously do localization and mapping of the environment using a camera and an attached IMU sensor. The method is called DM-VIO and can be used in self-driving cars, for example. DM-VIO is a project of the TUM Computer Vision Group led by our director Daniel Cremers.

Last but not least, we gave some insights into Stable Diffusion, a deep learning model that generates images from text descriptions. Stable Diffusion is developed by the LMU Computer Vision & Learning Group of our PI Björn Ommer. We explained how Stable Diffusion learns the connection between image and text through a latent diffusion model process.

©MCML

Poster 1: How does AI learn

AMELIE is the name of the museum exhibit that helped visitors understand machine learning principles.

©MCML

Poster 2: Reconstructing the world with AI

We presented DM-VIO, a monocular visual-inertial odometry system based on two novel techniques called delayed marginalization and pose graph bundle adjustment.

©MCML

Poster 3: DM-VIO demo

On this poster visitors saw how the Simultaneous Localization and Mapping unfolds

©MCML

Poster 4: Image generation using AI

Visitors saw how Stable Diffusion works and were able to try it out themselves.

22.01.2024