![EarthGenome](https://data.source.coop/earthgenome/earthindexembeddings/logo.png)
# EarthIndex Embeddings from Sentinel2 Temporal Mosaics 

[![CC-4 license](https://img.shields.io/badge/License-CC--4-blue.svg)](https://creativecommons.org/licenses/by/4.0/)
[![Sensor](https://img.shields.io/badge/Sensor%F0%9F%9B%B0%EF%B8%8F-Sentinel2-blue)](https://www.esa.int/Applications/Observing_the_Earth/Copernicus/Sentinel-2)
[![Format](https://img.shields.io/badge/Format-GeoParquet%20%F0%9F%8C%8E%EF%B8%8F-blue)](https://geoparquet.org/)

This repository holds embeddings generated the entire globe for the given date ranges.

This data is licensed under the [creative commons 4.0](https://creativecommons.org/licenses/by/4.0/) international license which, in summary, only requires attribution.

### Model
These embeddings are generated from the [SoftCon Model](https://github.com/zhu-xlab/softcon/tree/main) published by [Zhu Labs](https://www.asg.ed.tum.de/sipeo/home/)

### Naming
Artifacts are name according to their original UTM grid number, concatenated with the start and end dates.  For example `31TDF_2024-01-01_2025-01-01.parquet` will give you embeddings for Barcelona (31TDF) over the course of 2024.

### Artifacts and Metadata
All assets are Geoparquet in EPSG:4626 projection with point geometries. The fields are as follows:
- id: uint64
- embedding: float[384]
- geometry: Point in WKB


### Contact
Using this? Want work together? Ping us @ info@earthgenome.org or info@earthindex.ai