Lightnews — Scholar-powered news

Anton Obukhov @obukhov.ai · May 15

Original announcement of Marigold Depth (CVPR 2024 Oral, Best Paper Award Candidate): x.com/AntonObukhov...

Anton Obukhov @obukhov.ai · May 15

Team: Bingxin Ke (@bingxinke.bsky.social), Kevin Qu (@kevinqu.bsky.social), Tianfu Wang ( ), Nando Metzger (@nandometzger.bsky.social), Shengyu Huang, Bo Li, Anton Obukhov (@obukhov.ai), Konrad Schindler.
We thank @hf.co for their sustained support.

1 1

Anton Obukhov @obukhov.ai · May 15

A lot of cutting-edge research in related areas comes from ETH Zürich’s PRS lab, led by Prof. Konrad Schindler. Follow the lab for updates:
🐙 github.com/prs-eth
🤗 huggingface.co/prs-eth
🌐 prs.igp.ethz.ch
🎓 Open ETHZ BSc/MSc projects:
prs.igp.ethz.ch/education/op...

1 1

Anton Obukhov @obukhov.ai · May 15

🤗 Demo (Depth): huggingface.co/spaces/prs-e...
🤗 Demo (Normals): huggingface.co/spaces/prs-e...
🤗 Demo (Albedo): huggingface.co/spaces/prs-e...
🤗 Models: huggingface.co/collections/...

1 1

Anton Obukhov @obukhov.ai · May 15

🌎 Website: marigoldcomputervision.github.io
🧨 Tutorial: huggingface.co/docs/diffuse...
📗 Paper: arxiv.org/abs/2505.09358
🐙 Training code: github.com/prs-eth/Mari...

1 3

Anton Obukhov @obukhov.ai · May 15

New modalities include surface normals and intrinsic decompositions like albedo, material properties (roughness, metallicity), and lighting decompositions. Marigold proves to be an efficient fine-tuning protocol that generalizes across image analysis tasks.

1 6

Anton Obukhov @obukhov.ai · May 15

Big Marigold update!
Last year, we showed how to turn Stable Diffusion 2 into a SOTA depth estimator with a few synthetic samples and 2–3 days on just 1 GPU.
Today's release features:
🏎️ 1-step inference
🔢 New modalities
🫣 High resolution
🧨 Diffusers support
🕹️ New demos
🧶👇

1 8 45

Reposted by Anton Obukhov

Matteo Poggi @mattpoggi.bsky.social · May 14

🍸🍸The TRICKY25 challenge: "Monocular Depth from Images of Specular and Transparent Surfaces" is live! 🍸🍸 Hosted at the 3rd TRICKY workshop #ICCV2025, with exciting speakers! @obukhov.ai @taiyasaki.bsky.social

Site: sites.google.com/view/iccv25t...
Codalab: codalab.lisn.upsaclay.fr/competitions...

1 1

Anton Obukhov @obukhov.ai · May 1

@mattpoggi.bsky.social will probably know!

1 1

Anton Obukhov @obukhov.ai · Apr 25

The workshop report is now released! arxiv.org/abs/2504.17787

1 1 2

Anton Obukhov @obukhov.ai · Mar 23

Huawei Research Center Zürich is looking for a Research Scientist intern to work with me on advancing foundation models for computer vision, focusing on enhancing computational photography features in mobile phones. ˙✧˖°📸⋆｡˚

careers.huaweirc.ch/jobs/5702605...

Research Intern - Foundation Models for Computer Vision - Huawei Research Center Zürich

If you are enthusiastic in shaping Huawei’s European Research Institute together with a multicultural team of leading researchers, this is the right opportunity for you!

careers.huaweirc.ch

2 9

Anton Obukhov @obukhov.ai · Mar 14

Look at them stripes! A principled super-resolution drop by colleagues from PRS-ETH! Interactive demo with gradio-dualvision down in the post

Nando Metzger @nandometzger.bsky.social · Mar 14

We present Thera🔥: The new SOTA arbitrary-scale super-resolution method with built-in anti-aliasing. Our approach introduces Neural Heat Fields, which guarantee exact Gaussian filtering at any scale, enabling continuous image reconstruction without extra computational cost.

7

Anton Obukhov @obukhov.ai · Feb 28

bsky.app/profile/obuk...

Anton Obukhov @obukhov.ai · Dec 2

Introducing 🛹 RollingDepth 🛹 — a universal monocular depth estimator for arbitrarily long videos! Our paper, “Video Depth without Video Models,” delivers exactly that, setting new standards in temporal consistency. Check out more details in the thread 🧵

Anton Obukhov @obukhov.ai · Feb 28

RollingDepth rolls into Nashville for #CVPR2025! 🎸

1 4

Anton Obukhov @obukhov.ai · Feb 4

Previous posts: bsky.app/profile/obuk...

Anton Obukhov @obukhov.ai · Jan 31

Update about the 4th Monocular Depth Estimation Workshop at #CVPR2025:
🎉 Website is LIVE: jspenmar.github.io/MDEC/
🎉 Keynotes: Peter Wonka, Yiyi Liao, and Konrad Schindler
🎉 Challenge updates: new prediction types, baselines & metrics

Anton Obukhov @obukhov.ai · Feb 4

MDEC Challenge update! The 4th Monocular Depth Estimation Workshop at #CVPR2025 will be accepting submissions in two phases:
🚀 Dev phase: Feb 1 - Mar 1
🎯 Final phase: Mar 1 - Mar 21
Website: jspenmar.github.io/MDEC/
🌐 Codalab: codalab.lisn.upsaclay.fr/competitions...

Bring your best depth!

2 4 7

Anton Obukhov @obukhov.ai · Jan 31

Announcement: bsky.app/profile/obuk...

Anton Obukhov @obukhov.ai · Dec 21

The 4th Monocular Depth Estimation Challenge (MDEC) is coming to #CVPR2025, and I’m excited to join the org team! After 2024’s breakthroughs in monodepth driven by generative model advances in transformers and diffusion, this year's focus is on OOD generalization and evaluation.

Anton Obukhov @obukhov.ai · Jan 31

Update about the 4th Monocular Depth Estimation Workshop at #CVPR2025:
🎉 Website is LIVE: jspenmar.github.io/MDEC/
🎉 Keynotes: Peter Wonka, Yiyi Liao, and Konrad Schindler
🎉 Challenge updates: new prediction types, baselines & metrics

1 3

Anton Obukhov @obukhov.ai · Jan 25

Not finished of course, but it has been demonstrated that such massive compute might be excessive. Video makes sense, but top labs have already delivered astounding video generators with comparable resources.

1

Anton Obukhov @obukhov.ai · Jan 25

What's the next frontier after LLMs, that will demand nuclear-powered GPU clusters? No agents or AGI please

3 2

Anton Obukhov @obukhov.ai · Dec 21

Stay tuned for more updates and resources!

MDEC Team: Matteo Poggi (@mattpoggi.bsky.social), Fabio Tosi, Ripudaman Singh Arora, Anton Obukhov (@obukhov.ai), Jaime Spencer, Chris Russell (@cruss.bsky.social), Simon Hadfield, Richard Bowden.

1

Anton Obukhov @obukhov.ai · Dec 21

The 4th Monocular Depth Estimation Challenge (MDEC) is coming to #CVPR2025, and I’m excited to join the org team! After 2024’s breakthroughs in monodepth driven by generative model advances in transformers and diffusion, this year's focus is on OOD generalization and evaluation.

1 3 22

Reposted by Anton Obukhov

Nando Metzger @nandometzger.bsky.social · Dec 19

Monocular depth meets depth completion🚀 Check out our latest work where we modified Marigold to a zero-shot depth completion tool. Everything without retraining🌼 (This paper, for once, contains geese instead of cats😄 keep an eye open)

1 1 20

Anton Obukhov @obukhov.ai · Dec 19

Team ETH Zürich: Massimiliano Viola, Kevin Qu (@kevinqu.bsky.social), Nando Metzger (@nandometzger.bsky.social), Bingxin Ke (@bingxinke.bsky.social), Alexander Becker, Konrad Schindler, and Anton Obukhov (@obukhov.ai). We thank @hf.co and @gradio-hf.bsky.social for their continuous support.

3

Anton Obukhov @obukhov.ai · Dec 19

🌎 Website: marigolddepthcompletion.github.io
🤗 Demo: huggingface.co/spaces/prs-e...
📕 Paper: arxiv.org/abs/2412.13389
👾 Code: github.com/prs-eth/mari...

Marigold Depth Completion - a Hugging Face Space by prs-eth

Discover amazing ML apps made by the community

huggingface.co

1 4