Michael Oechsle

Research Scientist at Google in Zurich

[Michael Oechsle @mjoech]

Excited to present our project UNISURF at #ICCV2021 in Session 5A & 5B with @songyoupeng and @AutoVisionGroup.

Paper, Video, code, and more:
https://t.co/AqCrpouWpo

[Thomas Wimmer @wimmer_th]

Super excited to introduce

✨ AnyUp: Universal Feature Upsampling 🔎

Upsample any feature - really any feature - with the same upsampler, no need for cumbersome retraining.
SOTA feature upsampling results while being feature-agnostic at inference time.

[Dino Fuzi @DinoFuzi]

📢 Our paper CubeDiff has been accepted to ICLR 2025! 🎉🎉🎉

CubeDiff achieves SOTA panorama generation with only minimal architectural modifications to existing T2I diffusion models.

Check out our website @ https://t.co/Tntz4dtszy and paper @ https://t.co/JwwbmynsVD.

[Adam Kortylewski @AdamKortylewski]

Make sure to visit our workshop on "Generative Models for ComputerVision" on Tuesday.

We have an exciting speaker line-up, and 40 (!) posters on generative models.

@CVPR #CVPR2024 @fedassa @jiajunwu_cs @sainingxie @SiyuTang3 @Jamie_Shotton @YuilleAlan
K.Fragkiadaki A. Vedaldi

[Adam Kortylewski @AdamKortylewski]

Submit your short or long papers to our workshop on ""Generative Models for ComputerVision" @CVPR #CVPR2024

The deadline is March 27th.

We also have a fantastic speaker line-up!
@fedassa @jiajunwu_cs @sainingxie @SiyuTang3 @Jamie_Shotton @YuilleAlan @deviparikh K.Fragkiadaki

[Adam Kortylewski @AdamKortylewski]

Submissions are now open for our workshop on Generative Models for Computer Vision!
@CVPR #CVPR2024

We also have an initial line-up of keynote speakers.

https://t.co/2TGy7r24GT

@fnzhan0507 @LingjieLiu1 @Mi_Niemeyer @mjoech @YuilleAlan @VcaiMpi

[Michael Niemeyer @Mi_Niemeyer]

Recent reconstruction systems focus on either complex scenes (e.g. Neuralangelo) or shiny surfaces (e.g. Ref-NeRF). In UniSDF, we combine strengths from both worlds both leading to high-fidelity reconstructions of complex scenes with reflections: https://t.co/lFYRl1DQfJ

[Daniel Duckworth @duck]

Introducing SMERF: a streamable, memory-efficient method for real-time exploration of large, multi-room scenes on everyday devices. Our method brings the realism of Zip-NeRF to your phone or laptop!

Project page: https://t.co/WOzE4ApKAH
ArXiv: https://t.co/lugQXu3mQZ

(1/n)

[Christian Reiser @ChrisJReiser]

In less than an hour I am going to present our paper MERF at #SIGGRAPH2023 in Petree Hall D.

MERF allows you to interactively explore large scenes on a laptop in the browser. Check out our web demo: https://t.co/a4uOxriY1D

By the way we have also released the entire code now!

[Songyou Peng @songyoupeng]

🚨 Are neural implicit representations applicable for larger-scale SLAM? Check out our NICE-SLAM👍! #CVPR2022

NICE website: https://t.co/6JlKyGXKCH
NICE code: https://t.co/vRJPAh9I6v

NICE collaborations w/ Zihan Zhu (undergrad) @visionviktor @Martin_R_Oswald @mapo1 et al.

1/6

[Songyou Peng @songyoupeng]

🚨 Check out our @NeurIPSConf Oral paper: Shape As Points (SAP)!

We represent watertight meshes of any topology as a lightweight point cloud at speed. The key idea is to bridge them with a super simple, fast, versatile, and differentiable Poisson solver.

https://t.co/cKBHLFwNpg

[Katja Schwarz @K_S_Schwarz]

Do you ever wonder what's really going on in your GAN? Our NeurIPS'21 paper investigates the cause of frequency artifacts in generated images. Nice addition: we develop simple testbeds to analyze generators and discriminators individually.
https://t.co/HuYFzdrU2N

[Axel Sauer @AxSauer]

Thanks for sharing our newest work!

As a bonus, more of our generated Pokemon:

[Michael Oechsle @mjoech]

Our key insight is that implicit surface models and radiance fields can be formulated in a unified way, enabling both surface and volume rendering using the same model. This unified perspective enables reconstructing accurate surfaces without input masks.

[Michael Oechsle @mjoech]

Excited to present our project UNISURF at #ICCV2021 in Session 5A & 5B with @songyoupeng and @AutoVisionGroup.

Paper, Video, code, and more:
https://t.co/AqCrpouWpo

[Cyber Valley @Cyber_Valley]

We're excited about #ICCV2021! With 24 accepted papers, #CyberValley partner institutions continue to demonstrate their strength in #ComputerVision. https://t.co/OcMid7Swvj @ICCV_2021 @Michael_J_Black @MPI_IS @uni_tue

[AK @_akhaliq]

NEAT: Neural Attention Fields for End-to-End Autonomous Driving
pdf: https://t.co/kjTfnl2qR6
abs: https://t.co/l2SzAWD0iE
github: https://t.co/Et4EY7jkNo

[Christian Reiser @ChrisJReiser]

KiloNeRF was accepted at #ICCV2021!! By using thousands of tiny MLPs novel views can be rendered in real-time (50 fps) with NeRF's quality. Setting it apart and crucial for scalability, the repr. is decoded on the fly and so only needs <100MB of GPU mem.
https://t.co/6GZ4Jyk1CW

[Cyber Valley @Cyber_Valley]

#CyberValley shines at #CVPR2021. 27 papers accepted and four best paper nominations. @MPI_IS @uni_tue @Bosch_AI #AI #ComputerVision #AI https://t.co/FQtX67wAFC

[Jonas M. Kübler @jonas_kubler]

🧐 Can Quantum Machine Learning Models outperform classical ML models?

We worked on a few steps towards answering this question: https://t.co/kZiWiywTMM
with Simon Buchholz and @bschoelkopf

a Thread 📜 1/8

[Christian Reiser @ChrisJReiser]

We just published the code for KiloNeRF including pretrained models and an interactive viewer:

Code: https://t.co/TPRY5VHFy2
Paper: https://t.co/3PJ6eAs5mr

We further optimized the implementation and the Lego scene is now rendered at smooth 50 FPS!