dAIper-Hack
A hackathon to turn ideas into impact
- hands-on innovation event jointly organized by @fau.de and P&G
- three days
- modern and classical Computer Vision and AI approaches
- at the Scientific Center for AI and SuperTech
Apply now! daiper.tf.fau.eu
(latest by February 6th)
Second day at #automatica showcasing our KARVIMIO project
Had a great experience presenting our work on 3D scene reconstruction from a single image with @visionbernie.bsky.social at #3DV2025 🇸🇬
andreeadogaru.github.io/Gen3DSR
Reach out if you're interested in discussing our research or exploring international postdoc opportunities @fau.de
🤝 Shared first authorship with Nicolas Schischka @mertkaraoglu.bsky.social — huge thanks to the entire team for the incredible collaboration and hard work!
#ICRA2025 #RA_L #ComputerVision #NeRF #Robotics #CameraLocalization #DynaMoN #AI #MIRMI #TUM #FAU
🔍 DynaMoN tackles the challenge of robust and accurate camera localization in dynamic scenes, pushing the boundaries of Neural Radiance Fields in motion-rich environments.
🚀 Exciting news! Our paper DynaMoN: Motion-Aware Fast and Robust Camera Localization for Dynamic Neural Radiance Fields has been accepted to IEEE RA-L and will also be presented at #ICRA2025! 🎉
Gemeinsam gegen Blutkrebs: Registrieren und Leben retten!🩸
Ab nächster Woche gibt es an der FAU Registrierungsaktionen, um sich im Rahmen von #FrankenfuerdasLeben in die Datenbank der DKMS aufnehmen zu lassen. Kommt vorbei!
www.fau.de/2025/04/news...
I’ll be soon hiring new PhD students and a PostDoc in Aarhus —if you're interested in XR, perception, and visual computing please get in touch! I’m also at #chi2025 until the end of the week. #academia #phd #jopopening #denmark #XR #AR
Thank you ☺️
Finally an impactful conference deadline
First day of #IEEEVR 2025. starting early in the morning with the 3rd Workshop on XR Technologies for Healthcare, Wellbeing, and Medicine 2025. Presenting our review paper with a quick lighting talk and two poster session. #VR #workshop #conference
Congrats 🎉
🙏 Huge thanks to my amazing collaborators! It’s been incredible working on these projects, and I’m thrilled to see them make it to IEEE VR 2025! Looking forward to engaging discussions at the conference! 🚀🎉
Authors: Hannah Schieber, Luisa Theelke, Dirk Wilhelm, Cristina Piazza, Ulrich Eck, Julian Kreimeier, and Daniel Roth
🔬 This work explores the current state and future directions of XR-based teleconsultation for medical interventions.
🏥 Workshop Presentation
📄 Extended Reality Teleconsultation for Medical Interventions: State of the Art and Perspectives
🩺 To be presented at the 3rd Workshop on XR Technologies for Healthcare, Wellbeing, and Medicine 2025
✅ Comprehensive Evaluation: A detailed technical assessment of SCGS.
✅ User Study: A two-phase study on user experience and perception of SCGS
Our Contributions:
✅ SCGS: A state-of-the-art Semantic-Controlled Gaussian Splatting (SCGS) approach that surpasses existing methods.
✅ New Dataset: A challenging outdoor NVS dataset with semantic labels, now publicly available 🔗 Dataset.
🚀 Excited to present two works at IEEE VR 2025
📄 Semantics-Controlled Gaussian Splatting for Outdoor Scene Reconstruction and Rendering in Virtual Reality
Authors: Hannah Schieber, Jacob Young, @tobiaslanglotz.bsky.social , @stefaniezollmann.bsky.social, and Daniel Roth
youtu.be/57QtfvArSco
Congrats to Daniela et al.! Super happy that one of my drawings made it to the title page of the journal! And even more happy for Daniela and here co authors on their new research :) #FAU #drawing
Thanks! I saw the code is available so I will give it a shot probably when I have some free gpu space. Is it trainable on one single 24GB? Otherwise I might be limited from GPU perspective.
Really cool work! Have you tried it on some non typical datasets? With similar feature spaces ?
🎉🥳
👍🫶
Hannah Schieber, Jacob Young, Tobias Langlotz, Stefanie Zollmann, Daniel Roth
Semantics-Controlled Gaussian Splatting for Outdoor Scene Reconstruction and Rendering in Virtual Reality
https://arxiv.org/abs/2409.15959
probably Gaga would perform better when using our DPT labels. Will test this in the future. Looking forward to more benchmarking on semantic gaussian splatting! #GaussianSplatting #Outdoor #ComputerGraphics #SCGS
Ours (#SCGS):
Gaga rendering, but it seems that the segmentation is really difficult to keep consistent:
One advantage of #Gaga could be that first pure #GaussianSplatting (30k) is trained and then the masks are associated and then in a second training step the semantics are learned (10k) which is different to Gaussian Grouping or our SCGS. (4/n)
