Ozgur Turkoglu

Postdoctoral Researcher @agroscope.
PhD @eth_en, @EcoVisionETH.
#ComputerVision & #MachineLearning & #RemoteSensing.

[Ozgur Turkoglu @0ozgur0]

Our paper will appear in #NeurIPS2022. Deep Ensembles are powerful for accuracy boost and epistemic uncertainty quantification but computationally expensive. We make them very cheap by using multiple affine transformations in batchnorm layers @EcoVisionETH https://t.co/bPB7np25a4

[Anton Obukhov @AntonObukhov1]

Introducing Marigold 🌼 - a universal monocular depth estimator, delivering incredibly sharp predictions in the wild! Based on Stable Diffusion, it is trained with synthetic depth data only and excels in zero-shot adaptation to real-world imagery. Check it out:

🌐 Website: https://t.co/rBXhxQChn4
🤗 Hugging Face Space: https://t.co/CYvZZQtEYr
📄 Paper: https://t.co/eEw11yW6MY
👾 Code: https://t.co/0L49Znp2z1

The team: Bingxin Ke (@KBingxin), yours truly (@AntonObukhov1), Shengyu Huang (@ShengyHuang), Nando Metzger (@NandoMetzger), Rodrigo Caye Daudt (@rcdaudt), and Konrad Schindler.

#ComputerVision #PRS #ETHZurich

[Nando Metzger @NandoMetzger]

🛰️🌆 Exploring Urban Futures Through Satellite Imagery 🌆🛰️

Thrilled to announce our most recent publication in PFG, "Urban Change Forecasting from Satellite Images." This research delves into an intriguing question and offers a new approach.
https://t.co/zmqRvbFOLT

[Anton Obukhov @AntonObukhov1]

🎨 Excited to share our latest paper: "Breathing New Life into 3D Assets with Generative Repainting"! We present a novel approach to repaint 3D models using text prompts, pretrained Stable Diffusion, and NeRF. A step towards assisting content creators! 🖌️

📄 Paper: https://t.co/0deQ7lNLIE
🦄 Project Page: https://t.co/Hv4heyCEHz
👾 Code: https://t.co/pxXBfFsmuX
🤗 HuggingFace Space: https://t.co/FQY9YvtmkL

Thrilled to present this as a poster at the @ICCVConference workshop on AI in 3D Content Creation (AI3DCC) next week, and as an oral talk at @BMVCconf, alongside the first author, Tianfu. We're looking forward to engaging discussions. Come talk to us!

Big thanks to my amazing colleagues Tianfu Wang (@TianfuWang2), Menelaos Kanakis (@menelaoskanakis), Konrad Schindler, and Luc Van Gool.

#GenerativeArt #GenerativeAI #ComputerVision #ContentCreators #ICCV #ICCV2023 #BMVC #BMVC2023 #CVL #PRS #ETHZurich

[Nando Metzger @NandoMetzger]

Need some high-resolution population density maps for Tanzania🇹🇿, Rwanda🇷🇼, Uganda🇺🇬, Mozambique🇲🇿, or Zambia🇿🇲? I recently produced maps using POMELO and open-sourced them on our GitHub page. https://t.co/pkv8D4PGGf
Some teaser images⬇️🧵

[Nando Metzger @NandoMetzger]

POMELO is a method that helps improve the quality of existing population data by employing a weakly supervised regression system. In my latest blog post, I delve into the technical details of our "Nature, Scientific Reports" publication. (5min read😉)
https://t.co/IU5eiE9Zqv

[Charlotte Pelletier @CharlottePlltr]

🛰️🔍Looking for a satellite imagery dataset with a temporal dimension? 📈
Thanks to Corentin's comprehensive list🙌, you can now discover various pre-processed datasets with annotations to advance your methodological developments.
Check it out here: https://t.co/Q5ko022O4t

[Ozgur Turkoglu @0ozgur0]

Our paper: Pixel-based yield mapping and prediction from Sentinel-2 using spectral indices and neural networks got accepted to Field Crops Research: https://t.co/6ojr98MFOx. Code and dataset are available: https://t.co/MARRB87Ubp.

[Ozgur Turkoglu @0ozgur0]

i disagree. i think we should think about how/why undergrads can get in those venues (many of us care so much) so often?

[F. Güney @ftm_guney]

undergrads complaining that I’m not writing ref. letters for them even though they got an A in my course: impossible to get into a grad. program without any research experience, even papers. maybe in another field or by some other prof but not in ours, and definitely not by mine.

[Matthias Niessner @MattNiessner]

For invited workshop talks at CVPR, I would encourage everyone to invite more junior people.

There's so many rising stars who give amazing talks!

[Zan Gojcic @ZGojcic]

After seeing all the amazing results of 2D generative models, it is now time for 3D!

Check out our new #NeurIPS2022 work GET3D. A powerful 3D generative model that directly outputs high-quality 3D textured meshes. @NVIDIAAI #3D #DL #ML #CV

[Giovanni M Farinella @GMFarinella]

Thanks Davide Scamamuzza for the nice video!

The lectures of the 2022 edition of ICVSS are finished.

Thanks all speakers, sponsors and all the amazing students!

See you next edition!

PS: BTW continue to follow us on social media…there’ll be social activities tonight ;-)

[Ozgur Turkoglu @0ozgur0]

@SahilAnchal @MarcCoru @CharlottePlltr thank you:)

[Ozgur Turkoglu @0ozgur0]

I am pleased to give a talk at the ISPRS Congress 2022 Special Session on Satellite Image Time Series Analysis. I tried to draw attention to what I actually learned during my Ph.D. Thanks, @MarcCoru and @CharlottePlltr for inviting me! #ISPRS2022

Slides:
https://t.co/ArGXgQKpFp

[Ozgur Turkoglu @0ozgur0]

Check out our latest work: FiLM-Ensemble, a deep, implicit ensemble method based on the concept of Feature-wise Linear Modulation. It comes very close to the upper bound of an explicit ensemble of networks at a fraction of the memory cost. https://t.co/bPB7np1xkw
@EcoVisionETH

[ETH Zurich @ETH_en]

Happy #PrideMonth #LoveIsLove 🌈

[Eric Brachmann @eric_brachmann]

We offer visual #relocalisation at 30.000 sites world wide, with especially dense coverage of 6 launch cities. This scales up as we speak. Dense coverage in 100 cities by end of year, and we will continue beyond that with the help of our users.
#VPS @LightshipAR #LightshipSummit

[Nico Lang @nicolangnl]

We are attending @esa's #LPS22 in Bonn. Come and talk to my colleagues about #deeplearning for #crop mapping during the today's poster session, infos below. 👇@andrescamilor @0ozgur0 @DinoFuzi

[Erik Bekkers @erikjbekkers]

I'm making videos on group equivariant DL and am this content with the end of lecture 1 that I thought to share it! 😁 I'll put them out as soon as they're done!

Topics:
1 Regular G-CNNs
2 Steerable G-CNNs
3 Equivariant graph NNs
4 Geometric latent spaces

Watch this space!

[Eric Brachmann @eric_brachmann]

We uploaded alternative "ground truth" and full SfM models for the #relocalisation datasets 7Scenes and 12Scenes.

Working on reloc towards #ECCV2022 and having trouble beating SOTA? The "ground truth" might play a role.

https://t.co/suOPK28MmJ

#ICCV2021 #betterlatethannever