We’re excited to announce that our work “UnCRtainTS: Uncertainty Quantification for Cloud Removal in Optical Satellite Time Series” has been accepted at this year’s #CVPR2024 EarthVision ! i/iii

Change predictions based on my and Patrick Ebel 's method in Mariupol over the last month, which subsequently would require ground truth confirmations. This may be areas impacted by explosion, but also other changes. Note that the method does not use any post-event optical image.

In our paper, we curate a global and multi-decadal dataset of paired in-situ tide gage data, atmospheric reanalysis and ocean state simulations to enable training and benchmarking of deep neural networks.

The challenge we tackle is that of combining the accuracy of sparse in-situ data with the dense coverage of global atmosphere and ocean state products. While the latter are available worldwide, in-situ infrastructure is often limited for communities most at risk of such hazards.

Our approach aims to densify sparsely deployed but precise in-situ measurements on a global scale. This is accomplished by e.g. broadcasting forecasts from gauged to ungauged sites, input data dropout and auxiliary supervision.

Curious how clouds affect remote sensing applications, such as land cover classification? Jakob Gawlikowski, Michael Schmitt at UniBw M, Xiaoxiang ZHU and me provide an analysis and interpretation of the effects of cloud coverage!

Check out ieeexplore.ieee.org/document/99568…

Everyone, please check out our latest work on global forecasting of storm surges at ungauged locations, accepted at #CVPR2024 EarthVision 2024! arxiv.org/abs/2404.05758 @ESA ESA Earth Observation #Philab

Exciting news: Dr. Ebel Patrick Ebel has successfully defended his PhD on #cloudremoval in #AI4EO . Congrats, Patrick! 🎉 In my view, he did significant contributions towards #trustworthy and #scalable cloud removal in satellite images for real world applications.

We‘re organizing the joint ESA-ECMWF workshop in Frascati next year. Come and join us! Submissions are now open.

I and Patrick Ebel stressed the importance of change detection systems that can work without post-change optical data (e.g., ieeexplore.ieee.org/document/95383…). Ukraine crisis further confirms it, there is hardly any cloud-free optical S2 image there so far. #remotesensing #AI #disastermapping

Alpha Phenix Finance | Empowering the Future of DeFi World of Hedgehogs
عمرو عبد السلام
Patrick Ebel

LBank.com 똥코인 (DongCoin) pedro Patrick Ebel @owueb

Elber Nabuco Silveira De Carvalho 🔰 TheYellows 🔰 ..xiaoxuuu🎧🎶 Nmsl Dr. Ing. Thomas Duriez Norana ❛ # yani ; rpwp ꒱ ぱこさん 🩵نور🩵 ほげぇた Patrick Ebel ちろちゃん☆チロくま @biblewcp2512 Sunnysuz 🟧🇺🇸🇺🇦🌻 Abbas Akbar @omer_eren17 Betterkeepcalm G REX KABURU 🔰 TME 🔰

on a related matter, also see our Master student Ziqi’s work on “Explicit Haze & Cloud Removal for Global Land Cover Classification” presented earlier this year sites.google.com/view/rainfores…

Storm surges are a major hazard in the context of cyclone impact and land subsidence, which are prognosed to become more problematic due to climate change and rising sea levels.

UnCRtainTS combines cloud removal in optical satellites with pixel-wise uncertainty predictions. In practice, it can be used to control for reconstruction quality. You can find the preprint at arxiv.org/abs/2304.05464 iii/iii

Carlos Jiménez Gobierno de Atizapán de Zaragoza Fiscalía Edoméx Secretaría de Seguridad del Estado de México Y te faltaron otras muchas más...

Patrick Ebel #CVPR2024 EarthVision YES!

Patrick Ebel Robin Cole Thank you for the kind words Patrick Ebel - also, those interested in cloud removal problems should definitely check out Patrick’s excellent works like DSen2-CR, GLF-CR (among many others) as well as their related datasets!