Tim Gattinger

Interests in AI with focus on remote sensing. Works as engineer in a project monitoring forests with remote sensing and AI In Schleswig-Holstein (Germany).

[Tim Gattinger @GattingerTim]

Individual tree detection shows promising initial results. Tested over 392 km2 forest in Schleswig-Holstein. Aerial pictures and Laserscan form the basis, classification with UNET. Aim: Classification for all forest areas in SH
#SchleswigHolstein #KI #Forest #Fernerkundung

[Tim Gattinger @GattingerTim]

I had great entertainment and fun at linked in this morning 🤣 unfortunately it’s in German.. but feel free robuste Google translate or sth.. it’s worth it 😉 tells a lot about how a guy who thinks is on top of society sees poverty and social problems in Germany

[Tim Gattinger @GattingerTim]

Hier eine LinkedIn Perle 🤣 zum amüsieren.. https://t.co/3q9sCOl7dx

[Tim Gattinger @GattingerTim]

@elonmusk Elon, you suck at video games.. there are only fake accounts letting you pretend you’re good at it

[Robin Cole @robmarkcole]

🎙️I recorded podcast episode with Alastair Graham (former co host of Scene from Above) where we discussed the intersection of EO & deep learning, and dived into building the satellite-image-deep-learning community 👥🛰️

https://t.co/sCwYeaBg8w

[UNOSAT @UNOSAT]

🇹🇩 Satellite-detected flood impacts in N'Djamena City, Chad (26 Aug 2024): 50 km² of land affected, with around 7,900 buildings impacted.

📸https://t.co/5yYTdnbCks

#Chad #FloodMonitoring #Geospatial #Satellites #Flooding

[The Open Mapping Hub-Eastern & Southern Africa Hub @openmapping_esa]

Call for Mappers: We urgently need your assistance to map the recent floods in the DRC
https://t.co/uVzwM9PcVS
https://t.co/IgRSNM5dwx
https://t.co/IJk3USasij
https://t.co/4M8T34mudk
https://t.co/ZnBDkOPJZp

#DisasterResponse
#ESADisasterResponse
#DRCFloods
#DRCFloodResponse

[ipfconline @ipfconline1]

Fighting #deepfakes, shallowfakes and media manipulation

https://t.co/fw4en3mF2v via @techxplore_com @UofMaryland

#CyberSecurity #AI #AIEthics
Cc @HaroldSinnott @dinisguarda @Fisher85M @MarshaCollier @kalydeoo @Nicochan33

[Tim Gattinger @GattingerTim]

West Coast of Schleswig-Holstein at 5.5.2022. Satellite data from Sentinel-2 (Copernicus). True color and Infrared visualisation. #SchleswigHolstein #remotesensing #GIS #copernicusmarine #Earth #Science

[Sumanth @Sumanth_077]

Pandas is a powerful data analysis and manipulation library for Python!

NVIDIA just made Pandas 150x faster with zero code changes🔥

All you have to add is just a couple of lines of code:

%load_ext cudf.pandas
import pandas as pd

Thread🧵👇

[Tim Gattinger @GattingerTim]

@datenschatz Yes, there’s no reason why it shouldn’t work on urban areas as well! As long as the tree types in urban areas have been trained as well.

[Tim Gattinger @GattingerTim]

Individual tree detection shows promising initial results. Tested over 392 km2 forest in Schleswig-Holstein. Aerial pictures and Laserscan form the basis, classification with UNET. Aim: Classification for all forest areas in SH
#SchleswigHolstein #KI #Forest #Fernerkundung

[Tim Gattinger @GattingerTim]

@JensWiesehahn Yes it’s kind of a classical approach. Also a tree crown delineation algorithm in the background from Dalponte and Coomes (2016) (not depicted in the images). I guess the two step in one would work well, but there’s a need to generate well segmented train data too. Let’s Email!

[Tim Gattinger @GattingerTim]

@McCormicGEO Thanks Justine!

[Tim Gattinger @GattingerTim]

@BioMirela Thank you Mirela!

[Tim Gattinger @GattingerTim]

@MykolaKozyr Thank you! AFAIK it’s a key problem to have precise location of single tree data in order to validate the results. However, we are cooperating with forest authorities working on a more precise location of their already collected data we could use then for meaningful validation.

[Tim Gattinger @GattingerTim]

@JensWiesehahn ATM we‘re testing two steps 1) segmentation or ITD with Laser data 2) Classification with UNET

[Tim Gattinger @GattingerTim]

I published a app a year ago with Streamlit, Python over GitHub and Heroku. It shows river flood levels in Schleswig-Holstein. It shows the flooding in Baltic Sea 2023 was a one in hundred year event..

https://t.co/WxrWrdsGMT

#GIS #SchleswigHolstein #Python #GitHub

[Tim Gattinger @GattingerTim]

@jens_schwaab Das Haus ist hässlich wie die Nacht, da hilft auch der weiße Anstrich nicht.. und am besten alles zupflastern rundherum..

[Tim Gattinger @GattingerTim]

@William_Lidberg Segmentation with Machine Learning on Raster data (sklearn / tensorflow) maybe

[Tim Gattinger @GattingerTim]

@elonmusk Shut your mouth an pay taxes..