Kategorie: ‘Theses’
Data Curation for Decentralised Model Training using Representation Learning for Feature Embeddings
Performing Distributed Analytics on Decentralised Geo- and Hydrological Data
Model Training through Curiosity-based Latent Space Exploration on Decentralized Data
Development of Generative Models Trained on Decentralised Hydrological Data
Component Crawler for Mixed Reality Projects
For mixed reality applications, developers usually choose a modular development approach where the application-logic is segmented into separate components which only realize one feature. In well-designed projects, the components are reusable in different contexts within the project but also outside of it. Additionally, code from these projects can be used to find examples how to use APIs or libraries. However, many elementary features of mixed reality applications are re-implemented in new projects, instead of leveraging components from existing open-source projects.
Persistent Association of Virtual Objects in Mixed Reality
A mixed reality environment immerses users in a three-dimensional space. This allows them to organize digital content in the entire room, e.g. task cards in immersive project management. A natural way to sort related objects is by placing them close to each other. However, this kind of spatial co-location of objects is not enough to stress their relation to each other. If another user who is unaware of the association moves one of the objects, the spatial connection is lost. Hence, lines in 3D space are required which establish a visual and persistent connection between the objects. In a previous project, an initial system was implemented which uses straight lines to connect the objects. However, straight lines have some shortcomings since they intersect with obstacles between the two endpoints. A better connection system can use 3D curves to avoid the obstacles in an efficient way.