iiinova — where your digital self keeps evolving

Reposted by Maximilian Weiherer

VisionBernie - Bernhard Egger @visionbernie.bsky.social · 11/09/2026

"I'm looking for a postdoc" "3D Computer Vision" 🎤🫳 #ECCV2026 And yes, I really am looking for a postdoc @csprofkgd.bsky.social please spread the world

4176

Maximilian Weiherer @maxweiherer.bsky.social · 11/09/2026

(5/5) This is joint work with Chukwudi Williams Umah and @visionbernie.bsky.social! Project page: mweiherer.github.io/vksr/

010

Maximilian Weiherer @maxweiherer.bsky.social · 11/09/2026

(4/5) At the same time, VKSR maintains strong performance in a sparse setting, unlike widely used Screened Poisson Surface Reconstruction.

110

Maximilian Weiherer @maxweiherer.bsky.social · 11/09/2026

(3/5) This naturally (without any engineering tricks) makes kernel surface reconstruction entirely local, allowing us to reconstruct millions of points without subdivision schemes and with speedups of up to 180 times compared to existing low-rank-based methods.

110

Maximilian Weiherer @maxweiherer.bsky.social · 11/09/2026

(2/5) VKSR exploits the connection between kernel surface reconstruction and Gaussian Processes (GPs), and replaces suboptimal low-rank approximations with an approximation technique from the GP literature: Vecchia's approximation.

110

Maximilian Weiherer @maxweiherer.bsky.social · 11/09/2026

Introducing Vecchia Kernel Surface Reconstruction (VKSR), an implicit surface reconstruction method that efficiently scales kernel surface reconstruction to dense and large-scale point clouds with millions of points. I'm presenting VKSR at #ECCV2026 in Malmö now at poster #190!

131

Reposted by Maximilian Weiherer

Lukas Meyer @meyels.bsky.social · 03/09/2025

🚀 Excited to share that our paper "Towards Integrating Multi-Spectral Imaging with Gaussian Splatting" was accepted at VMV 2025! 🎉 🔗 Project: meyerls.github.io/towards_mult... 📄 Paper: arxiv.org/pdf/2509.00989 💻 Code: github.com/j-gruen/MS-S... #VMV25 #GaussianSplatting #NeRF #Multispectral

173

Maximilian Weiherer @maxweiherer.bsky.social · 01/04/2025

I still remember the moment you first had this idea—during lunch at York University about 1.5 years ago. These guys have been working on this for more than ONE YEAR!!! This is insane, and way beyond a normal Sigbovik paper. Congrats on this huge success and an amazing paper!

040

Reposted by Maximilian Weiherer

VisionBernie - Bernhard Egger @visionbernie.bsky.social · 01/04/2025

Tired of seeing your papers being rejected from @cvprconference.bsky.social ? Let's revisit our 2023 #SIGBOVIK paper: "From Zero to Hero: Convincing with Extremely Complicated Math", led by @maxweiherer.bsky.social Paper: mweiherer.github.io/zero2hero.pdf Code: github.com/mweiherer/ze... 1/4

181

Maximilian Weiherer @maxweiherer.bsky.social · 20/02/2025

Absolut. Und dann auch noch j = sqrt(-1).

000

Maximilian Weiherer @maxweiherer.bsky.social · 20/02/2025

Jetzt weißt du wieso ich mich angegriffen fühle wenn mich jemand als Ingenieur bezeichnet. Ich möchte freundlich darauf hinweisen dass nicht alle Ingenieure pi auf 3 abrunden, oder e auf 2.

100

Maximilian Weiherer @maxweiherer.bsky.social · 19/12/2024

(5/5) Huge thanks to my co-authors, Antonia von Riedheim, Vanessa Brébant, @visionbernie.bsky.social, and Christoph Palm.

030

Maximilian Weiherer @maxweiherer.bsky.social · 19/12/2024

(4/5) The resulting model is free of correspondence errors, captures detailed surface geometry, and outperforms the RBSM in various surface reconstruction tasks. The iRBSM is publicly available for research purposes! Model: rbsm.re-mic.de/implicit Code: github.com/mweiherer/ir...

120

Maximilian Weiherer @maxweiherer.bsky.social · 19/12/2024

(3/5) Using an implicit (correspondence-free) representation to model breast shapes from surface-only 3D scans is highly advantageous, as registering these scans is extremely hard: almost no features, large deformations, and severe occlusions (esp. underbust).

120

Maximilian Weiherer @maxweiherer.bsky.social · 19/12/2024

(2/5) Our model uses a DeepSDF architecture and is trained directly on 168 raw 3D breast scans--without ground truth supervision. Contrary to the RBSM, no surface registration is required to bring training data into correspondence.

110

Maximilian Weiherer @maxweiherer.bsky.social · 19/12/2024

Happy to share our latest 3D generative breast model: the *implicit* RBSM, or iRBSM for short. As opposed to its PCA-based predecessor, the iRBSM leverages implicit neural representations, yielding a highly detailed and expressive 3D breast model. Paper: arxiv.org/abs/2412.13244

1103

Reposted by Maximilian Weiherer

Hannah Schieber @hannahhaensen.bsky.social · 09/12/2024

I am very excited to announce our paper “DynaMoN: Motion-Aware Fast and Robust Camera Localization for Dynamic Neural Radiance Fields” has been accepted to #IEEE RA-L. Paper: ieeexplore.ieee.org/document/107... Project page: hannahhaensen.github.io/DynaMoN/ Code: github.com/HannahHaense...

1103

Maximilian Weiherer @maxweiherer.bsky.social · 29/11/2024

The coffee machine in our lab broke, and won’t be fixed by the end of January 2025. This is just so unfair. I’ll better stay at home for the next two months.

060

Reposted by Maximilian Weiherer

Andreea Ardelean @andreead-a.bsky.social · 19/11/2024

Excited to share our paper which will be presented at #3DV2025 ✨ Gen3DSR: Generalizable 3D Scene Reconstruction via Divide and Conquer from a Single View ✨ 🌐 Project page: andreeadogaru.github.io/Gen3DSR 📄 Paper: arxiv.org/abs/2404.03421 👩‍💻 Code: github.com/AndreeaDogar... (1/5)

1236