Starting a blog post trilogy on 'GeoTIFFs to GPUs'! If you've struggled with using GDAL effectively in GPU workflows (read: machine learning pipelines), and/or have messed around with converting to other file formats because you thought 'GeoTIFF is slow', know that it doesn't have to be this way!

I'm working towards GPU-based decoding of Cloud-optimized GeoTIFFs into CUDA/GPU memory via Rust/FFI. Eventual goal is to do zero-copy to CuPy arrays on the Python side via DLPack. It'll be a few months of work, but let's start from part one:

https://weiji14.xyz/blog/geotiffs-to-gpus-part-1:-an-unexpected-conflict-multi-threaded-reads-too-many-jobs/

#GeoTIFF #CloudOptimizedGeoTIFF #nvTIFF #cog3pio #gischat #GeoML

Presenting my talk today on "The ecosystem of geospatial machine learning tools in the Pangeo world"

Slides: https://hackmd.io/@weiji14/foss4g2023oceania
Code: https://github.com/weiji14/foss4g2023oceania

#FOSS4GOceania2023 #FOSS4G #Pangeo #GeoML @zarr #xarray #xbatcher #zen3geo #gischat

☯️ #zen3geo v0.6.2. Tiny patch release with another STAC DataPipe! Release notes are at https://zen3geo.readthedocs.io/en/v0.6.2/changelog.html

Have you got a need to connect Spatiotemporal Asset Catalogs (STAC) to deep learning workflows? Or interested in cloud-native geospatial machine learning? I'd love to hear more from you at
https://github.com/weiji14/zen3geo/discussions/48, or just comment below!

@pytorch #TorchData #gischat #STAC #CloudNativeGeospatial #GeoML