Ph.D. Student in Computer Vision at TUM/DLR. I like to tackle global problems with satellite images
Joined October 2018
- Tweets79
- Following482
- Followers336
- Likes335
Lukas Kondmann retweeted
🚨1 WEEK LEFT🚨 In-person applications for @ClimateChangeAI’s second summer school are closing on Jan 31. Don’t miss this opportunity! (Applications for the virtual program will remain open)
More details here: climatechange.ai/events/summ…
Excited to announce the second @ClimateChangeAI Summer School (Jun-Aug 2023)!👩🎓☀️
Are you an #AI or #DataScience expert who wants to tackle #ClimateChange? Are you a #climate expert trying to use #ML in your work?
Apply to our virtual & in-person programs: climatechange.ai/events/summ…
Lukas Kondmann retweeted
You can hire me! 👇
IIES Job Market Candidate Tillmann von Carnap’s @TCarnap job market paper asks how rural marketplaces shape their surroundings and how this age-old way of trading can be leveraged for development. 1/5 bit.ly/3X3XiLx
Lukas Kondmann retweeted
JMLR has recently launched a Special Issue on ML for addressing problems of climate change! We welcome all submissions which use machine learning to address problems of climate change, including mitigation, adaptation, and climate science. [1/4]
Lukas Kondmann retweeted
I'm moving to Boston! 🎉🚀
I am incredibly excited to start as a postdoctoral researcher at @MSFTResearch in the New England (@MSRNE) lab in fall. I'll continue to work on geospatial problems, improving our cities & tackling climate change with machine learning.🤖🛰️🌏
Lukas Kondmann retweeted
Super excited to present our work @CVPR today afternoon! Join us at the poster 237 from 14.30-17.00 🛰🗺🌏
Our work DynamicEarthNet: Daily Multi-Spectral Satellite Dataset for Semantic Change Segmentation will be presented at #CVPR2022!
➡️ arxiv.org/abs/2203.12560
Tackling climate change is the defining task of our generation, so what part can the vision community play in that? 🧵⬇️
Lukas Kondmann retweeted
What happens if you train a Vision Transformer with a self-supervised dense clustering task?
In our #CVPR2022 paper we find that our method, Leopart 🐆, learns expressive object part embeddings.
Paper: arxiv.org/abs/2204.13101
Code: github.com/MkuuWaUjinga/leop…
🧵1/n
Lukas Kondmann retweeted
Interested in land use dynamics on the Earth's surface? Check our #DynamicEarthNet dataset -- daily, multi-spectral satellite observations of 75 AOIs across the globe with imagery from #Planetscope and #Sentinels, paired with high quality #LULC labels #AI4EO #CVPR
Lukas Kondmann retweeted
I will be giving a public talk about some of my research tomorrow as part of the QUVA Deep Vision lecture.
It'll be about self-supervised learning and privacy/ethics in CV (SeLa, PASS, SeLaVi, GDT and our GPT-2 bias paper).
Tune in here: meetup.com/Deep-Learning-AI/…
Lukas Kondmann retweeted
Interested in joining TUM and working on #photogrammetry and #remotesening, here is the great opportunity. A full professorship is open for application now:
portal.mytum.de/jobs/profess…
Lukas Kondmann retweeted
I am on the job market for summer / fall 2022, looking for academic (postdoc & faculty) & industry research positions focusing on #ML for #geospatial data, sustainability & #urban analytics.🛰️🌆🌏🤖
You can find my CV & more info on my website: konstantinklemmer.github.io/
Lukas Kondmann retweeted
Today we @Oxford_VGG release the dataset 𝐏𝐀𝐒𝐒: 1.4M, CC-BY licensed images that do not depict humans, human parts or other PID. It can be used for self-supervised pretraining and works as well as ImageNet.
info & models: github.com/yukimasano/PASS/
paper: arxiv.org/abs/2109.13228
PASS: An ImageNet replacement for self-supervised pretraining without humans
pdf: arxiv.org/pdf/2109.13228.pdf
abs: arxiv.org/abs/2109.13228
Lukas Kondmann retweeted
@kondmann presented "Under the Radar - Auditing Fairness in ML for Humanitarian Mapping". He compared the poverty estimated through electricity and the actual poverty population. Significant underestimations are found for scheduled tribes. #DatadrivenHumanitarianMapping #KDD2021
Lukas Kondmann retweeted
We are beginning preparations for #FAccT2022 — we are aiming to have it as a hybrid (virtual and face-to-face) conference in South Korea in June 2022, with a possible submission deadline in November 2021. Please stay tuned for more.
Lukas Kondmann retweeted
The 2nd track "Weakly-Supervised Multi-Class Change Detection" of DynamicEarthNet just started! Don't miss it: classic.grss-ieee.org/earthv…
@planetlabs @TU_Muenchen @DLR_en
@xiaoxiang_zhu @lealtaixe
Lukas Kondmann retweeted
❗️ Join the #DynamicEarthNet challenge with MUDS PIs @xiaoxiang_zhu, @lealtaixe and MUDS students (@kondmann and more) as part of the organization team 😎
👉 classic.grss-ieee.org/earthv…
Lukas Kondmann retweeted
Interested in #ChangeDetection? Don's miss out this year's @CVPR @EarthVisionWS challenge - #DynamicEarthNet. Time series data from @planetlabs and @ESA_EO's sentinel-2. Hand labelled LULC changes, funded by @DLR_de @BMWi_Bund
Lukas Kondmann retweeted
Looking for a motivated individual to team up with me as my deputy at TUM-SiPEO.
What we offer: to co-organize my lab consisting of a young and motived team on #AI4EO for #socialgood, while still being able to carry out independent research.
portal.mytum.de/jobs/wissens…
pls RT
Lukas Kondmann retweeted
Really happy to share our new work on arXiv! do check out our interactive cluster visualization here: robots.ox.ac.uk/~vgg/researc…
Check out our new work "Labelling unlabelled videos from scratch with multi-modal self-supervision" by @y_m_asano @mandelapatrick_ @chrirupp and Andrea Vedaldi in colab with @facebookai!
See below for our automatically discovered clusters on VGG-Sound! robots.ox.ac.uk/~vgg/researc…
Lukas Kondmann retweeted
@sherwang and I are very happy to present our work today on
Meta-Learning for Few-Shot Land Cover Classification
at the #CVPR2020 @EarthVision13 #Earthvision2020 workshop. @TU_Muenchen-#lmf + @Stanford @DavidBLobell
paper: arxiv.org/abs/2004.13390
video: youtu.be/aeBuLrP9ue4