Visual-AI-Server

• Subject Area: Computer Science
• Term: Funded in 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 562946811

A “Visual AI server” is required for computer science research, including methods for processing and generating predominantly visual data (images, videos, 3D/4D models). The focus here is on training and inference on very large amounts of data, on large-scale image processing for machine vision, as well as in the generation of synthetic visual data, e.g. by means of ray tracing and physically motivated simulation and generative AI models. The massive parallelization capability and the hardware architecture of GPUs (graphics cards) adapted to visual data and neural networks in a server with sufficient memory and CPU power to supply the GPUs are particularly suitable for this.

• DFG Programme: Major Research Instrumentation
• Major Instrumentation: Visual-AI-Server
• Instrumentation Group: 7000 Datenverarbeitungsanlagen, zentrale Rechenanlagen
• Applicant Institution: Christian-Albrechts-Universität zu Kiel
• Leader: Professor Dr.-Ing. Kevin Köser