Andreas Lugmayr

PhD Student in Computer Vision at ETH Zurich

[Anton Obukhov @AntonObukhov1]

Marigold V2 is out! (to appear at SIGGRAPH Asia 2026)

If you missed V1: Marigold post-trains an image generator into a depth estimator on one GPU -- the accessible research game.

V2 upgrades to a diffusion transformer: produces very sharp edges, and is quite versatile!

[Songyou Peng @songyoupeng]

Yay, finally! Introducing Vision Banana🍌 from @GoogleDeepMind, our unified model that outperforms SoTA specialist models on various vision tasks!

By treating 2D/3D vision tasks as image generation, we unlock a new foundation for CV.

Project page: https://t.co/GQgRi6mWwC

(1/5)

[Litu Rout @litu_rout_]

Continuous diffusion had a good run—now it’s time for Discrete diffusion!

Introducing Anchored Posterior Sampling (APS)

APS outperforms discrete and continuous baselines in terms of performance & scaling on inverse problems, stylization, and text-guided editing.

[Andreas Lugmayr @AndreasLugmayr]

Call for PhD research interns at Google:

Conduct research on Generative AI (Diffusion Models, Large Language Models) for commerce.

Deadline: Feb 15

Details are in the application form: https://t.co/hbhGWr4QGl

[Andreas Lugmayr @AndreasLugmayr]

Thanks to @unixpickle for noting that we missed out on other inpainting diffusion works in our preprint. We corrected that and apologize for this mistake.

[Andreas Lugmayr @AndreasLugmayr]

Inpainting with Diffusion Models - CVPR 2022

Inpainting approaches using unconditional diffusion model often suffer boundary artifacts.

Our inference schedule alleviates this by sampling every iteration multiple times.

Code + Models + Paper: https://t.co/Pno4odWWqV

[Andreas Lugmayr @AndreasLugmayr]

The blue regions are unknown and get inpainted by RePaint. The video shows each step of the inference. The pulsation comes from redoing the denoising.

Details: https://t.co/Pno4odWWqV

Thanks to my coauthors @MDanelljan, @afromero, @DrFisherYu, Radu Timofte and Luc Van Gool!

[Andrés Romero @afromero]

We are organizing a Challenge for Image Inpainting in the NTIRE 2022 Workshop that will be held on June in conjunction with CVPR 2022. Two cool things: a lot of different and challenging masks, and inpainting towards art creation. Repo: https://t.co/gmh9H3M0F7.

[Andreas Lugmayr @AndreasLugmayr]

CVPR 2022: Call for Workshop Paper

Sign up: https://t.co/4muRowrALX

[Prune Truong @prunetruong]

📯 ICCV 2021 Oral: "Warp Consistency for Unsupervised Learning of Dense Correspondences", with @MDanelljan, @DrFisherYu and Luc Van Gool.

Join us Session 8, 13th 9AM and 15th 4PM (EDT) if you want to know more!

paper: https://t.co/FAOPl9MIg3
video: https://t.co/N3AZq9XHoC

[Andreas Lugmayr @AndreasLugmayr]

WACV decisions are out!

[Anton Obukhov @AntonObukhov1]

Our #ICCV2021 paper on Adaptive Task-Relational Context, leveraging neural architecture search and attention mechanisms in the multi-task learning setting, is now paired with source code! @menelaoskanakis @stam_g

paper: https://t.co/JQDOGiT2YV
code: https://t.co/JrlNmmrThx

[Miguel Angel Bautista @itsbautistam]

Introducing Generative Scene Networks (GSN), a generative model for learning radiance fields for realistic scenes. With GSN we can sample scenes from the learned prior and move through them with a freely moving camera.
Arxiv: https://t.co/rYiFH4uhLp
Scenes sampled from the prior:

[Andreas Lugmayr @AndreasLugmayr]

@HahnerMartin Congrats!

[Martin Hahner @HahnerMartin]

Our paper "Fog Simulation on Real LiDAR Point Clouds for 3D Object Detection in Adverse Weather" has been accepted to ICCV 2021! Our preprint is available at https://t.co/ES0CYRaQZM. Code available at https://t.co/MylYEWbKi6.

[Andreas Lugmayr @AndreasLugmayr]

Great work @va_wolf!

[Martin Danelljan @MDanelljan]

Code released for our #CVPR2021 oral paper "DeFlow": A method for unpaired learning of image-to-image translation with normalizing flows.

Excellent master's thesis by @va_wolf👏
Also with @AndreasLugmayr et al.

arXiv: https://t.co/Owzw5qK8ye
code: https://t.co/RiY7HzwlDA

[Daniel Gehrig @DanielGehrig6]

Very excited to share this work! It really shows that we can do amazing things with event cameras.

[Andreas Lugmayr @AndreasLugmayr]

Find out more in their Q&A on Tuesday 10pm EDT

[Prune Truong @prunetruong]

Looking for dense and trustworthy correspondences?
Check out our #CVPR2021 oral paper w/ @MDanelljan:
Learning Accurate Dense Correspondences and When to Trust Them

Arxiv: https://t.co/80FFMo904w
Code: https://t.co/G1C3nc4PXw
Project page: https://t.co/Ee27975YrM

[Andreas Lugmayr @AndreasLugmayr]

@prunetruong @MDanelljan Congrats to the oral!

[Andreas Lugmayr @AndreasLugmayr]

Thanks to @deviparikh for this handy review organization spreadsheet and the guide on how to use it.

[Spreadsheet] https://t.co/nPsurlG5ZM
[Guide] https://t.co/U8IPdvZcKo