Proud to release the WorldStrat dataset with @ivanorsolic @alkalait and @esa Φ-lab: - ~10,000 km² of high-res SPOT satellite imagery - Multi-temporal matched low-res Sentinel2 images - Stratified wrt global land-uses - Enriched with sites under-represented in ML 🧵1/9
🌍 Download the dataset: https://t.co/tFDrOvwsol 📜 Paper: https://t.co/zFPxreHCAW ⌨️ Code: https://t.co/6yJdKQUbHz 📈 PWC: https://t.co/72EgPt6PRj 📄License: CC BY and CC BY-NC. 🧵2/9
zenodo.org/record/6810792
arxiv.org/abs/2207.06418
github.com/worldstrat/wor…
paperswithcode.com/dataset/worlds…
Each high-resolution @Airbus SPOT image (1.5 m/pixel) comes with multiple temporally-matched low-resolution images from the freely accessible lower-resolution Sentinel-2 satellites (10 m/pixel). 🧵3/9
We accompany this dataset with article, datasheet, & open-source Python package to: - rebuild or extend the WorldStrat dataset - train and infer baseline algorithms - and learn with abundant tutorials all compatible with the popular @sentinelhub EO-learn toolbox. 🧵4/9
We hope to foster broad-spectrum applications of ML to satellite imagery, and develop the same power of analysis from free public low-res Sentinel2 imagery as allowed by costly high-res imagery. 🧵5/9
We illustrate this specific point by training and releasing several highly compute-efficient baselines on the task of Multi-Frame Super-Resolution with cheap compute! Code and notebook tutorials available at https://t.co/6yJdKQUbHz. BSD-licensed pre-trained models too! 🧵6/9
github.com/worldstrat/wor…
With many thanks to @esa, its Φ-lab, Pierre Philippe Mathieu and Nicolas Longepe for their support and belief in this project, Grega Milcinski (@sinergise) and @sentinel_hub for taking us on board QueryPlanet. 🧵7/9
Extremely grateful to - @JamonVDH for his expertise on GHSL and UNHCR POIs - @AI_Micah for POIs of humanitarian interest - Josh Lyons (ex-@hrw) for consulting with us to reduce the risk of harmful uses - Moritz Besser (dida Datenschmiede GmbH) for their ASMSpotter data 🧵8/9
And thanks to @PeggyFischeresa and Bryan Keary, at @esa Third Party Missions programme, and @AirbusSpace for their support in making the imagery available publicly. 🧵9/9
