dc.contributor |
Háskóli Íslands |
dc.contributor |
University of Iceland |
dc.contributor.author |
Baker, Dirk Norbert |
dc.contributor.author |
Bauer, Felix Maximilian |
dc.contributor.author |
Schnepf, Andrea |
dc.contributor.author |
Scharr, Hanno |
dc.contributor.author |
Riedel, Morris |
dc.contributor.author |
Göbbert, Jens Henrik |
dc.contributor.author |
Hvannberg, Ebba Thora |
dc.date.accessioned |
2024-06-03T09:42:38Z |
dc.date.available |
2024-06-03T09:42:38Z |
dc.date.issued |
2024 |
dc.identifier.uri |
https://hdl.handle.net/20.500.11815/4936 |
dc.description.abstract |
E-infrastructures deliver basic supercomputing and storage capabilities but can benefit from innovative higher-level services that enable use-cases in critical domains, such as environmental and agricultural science.
This work describes methods to distribute virtual scenes to the GPU nodes of a modular supercomputer for data generation.
High information density virtual scenes, containing $>100$k geometries, typically cannot be rendered in real-time without techniques that change the information content, such as level-of-detail or culling approaches.
Our work enables the concurrent and partitioned coupling to the image analysis in such a way that the data generation is dynamic and can be allocated to GPU nodes on demand, resulting in the possibility of moving through a continuous virtual scene rendered on multiple nodes.
Within agricultural data analysis, the approach is especially impactful as virtual fields contain many individual geometries that coexist in one continuous system.
Our work facilitates the generation of high-quality image data sets which has the potential to solve the challenge of scarcity of well-annotated data in agricultural science.
We use real-time communication standards to couple the data production with the image analysis training.
We demonstrate how the use-case rendering impacts effective use of the compute nodes and furthermore develop techniques to distribute the workload to improve the data production. |
dc.description.sponsorship |
The German government funding to the Gauss Centre for Supercomputing via the InHPC-DE project (01—H17001).
This work has partly been funded by the EUROCC2 project funded by the European High-Performance Computing Joint Undertaking (JU) and EU/EEA states under grant agreement No 10110 1903.
This work has partly been funded by the German Research Foundation under Germany's Excellence Strategy, EXC-2070 - 390732324 - PhenoRob and by the German Federal Ministry of Education and Research (BMBF) in the framework of the funding initiative “Plant roots and soil ecosystems, significance of the rhizosphere for the bio-economy” (Rhizo4Bio), subproject CROP (ref. FKZ 031B0909A). |
dc.language.iso |
en |
dc.publisher |
Springer |
dc.relation.ispartofseries |
Nordic e-Infrastructure Collaboration (NeIC) Conference;2024 |
dc.rights |
info:eu-repo/semantics/openAccess |
dc.subject |
Computer Vision |
dc.subject |
Synthetic Data |
dc.subject |
Distributed Systems |
dc.subject |
HPC |
dc.subject |
Visualization |
dc.subject |
Sjónskynjun |
dc.title |
Adapting Agricultural Virtual Environments in Game Engines to Improve HPC Accessibility |
dc.type |
info:eu-repo/semantics/conferenceObject |
dc.description.version |
Pre-print (óritrýnt handrit) |
dc.identifier.journal |
Communications in Computer and Information Science |
dc.contributor.department |
Iðnaðarverkfræði-, vélaverkfræði- og tölvunarfræðideild (HÍ) |
dc.contributor.department |
Faculty of Industrial Eng., Mechanical Eng. and Computer Science (UI) |
dc.contributor.school |
Verkfræði- og náttúruvísindasvið (HÍ) |
dc.contributor.school |
School of Engineering and Natural Sciences (UI) |