SIH/Greater-Sydney-Trees-Buildings

---
license: mit
task_categories:
- image-segmentation
tags:
- geospatial
- map
- sydney
- trees
- buildings
pretty_name: Greater Sydney Tree and Building Footprints
size_categories:
- 1M<n<10M
---
Explore our Greater Sydney Building Footprint and Tree Patch datasets using a [PMTiles map here](https://pmtiles.io/#url=https%3A%2F%2Fhuggingface.co%2Fdatasets%2FSIH%2FGreater-Sydney-Trees-Buildings%2Fresolve%2Fmain%2Fgreater-sydney-trees-buildings.pmtiles&map=9.91/-33.8494/150.9332&showMetadata=true)


# aigis
<hr>
<p align="center">
<h1 align="center"><img src="images/aigis.png" alt="aigis logo" width=150px> <br> AIGIS</h1>

</p>
<p align="center">
<em>AI annotation, segmentation, and conversion tools for GIS imagery</em>
</p>

`aigis` is a comprehensive toolkit for aerial and satellite imagery acquisition, processing, annotation, and analysis using artificial intelligence. This repository contains three main components:

1. **annotate:** Tools for annotating aerial imagery data.
2. **convert:** Utilities for converting aerial imagery data to various formats, including COCO, GeoJSON, etc.
3. **segment:** Scripts and notebooks for segmenting aerial imagery using deep learning models.

## `aigis` datasets

Explore our two featured example datasets using a [PMTiles map here](https://pmtiles.io/#url=https%3A%2F%2Fhuggingface.co%2Fdatasets%2FSIH%2FGreater-Sydney-Trees-Buildings%2Fresolve%2Fmain%2Fgreater-sydney-trees-buildings.pmtiles&map=9.91/-33.8494/150.9332&showMetadata=true)

They are available as geoparquet and as PMTiles in this dataset.

### Building Segmentation

We've used the full `aigis` toolkit to fine tune segmentation models to detect and segment building outlines in aerial imagery. Our fine tuned model was then run on 37,000 high reslution aerial images across the entire Greater Sydney Urban area in New South Wales, Australia.

Our models's predictions are available as a shapefile in GeoParquet, with almost one million buildings.


### Tree Segmentation

We've used the full `aigis` toolkit to fine tune segmentation models to detect and segment tree patch outlines in aerial imagery. Our fine tuned model was then run on 37,000 high reslution aerial images across the entire Greater Sydney Urban area in New South Wales, Australia.

Our models's predictions are available as a shapefile in GeoParquet, with 2 million tree patches.


## About Us

This software was developed by the Sydney Informatics Hub, a core research facility of the University of Sydney.

The project team includes:

- Henry Lydecker
- Sahand Vahidnia
- Xinwei Luo
- Thomas Mauch
- Gordon McDonald

## License

This project is protected under the [MIT Licence](https://choosealicense.com/licenses) License. For more details, refer to the [LICENSE](https://choosealicense.com/licenses/) file.

## Acknowledgments

Acknowledgements are an important way for us to demonstrate the value we bring to your research. Your research outcomes are vital for ongoing funding of the Sydney Informatics Hub. If you make use of this software for your research project, please include the following acknowledgement:

>This research was supported by the Sydney Informatics Hub, a Core Research Facility of the University of Sydney.