https://github.com/pierresegonne/oil-riggery
https://github.com/pierresegonne/oil-riggery
Last synced: 3 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/pierresegonne/oil-riggery
- Owner: pierresegonne
- License: mit
- Created: 2023-07-23T14:46:35.000Z (over 2 years ago)
- Default Branch: main
- Last Pushed: 2023-10-22T10:23:50.000Z (about 2 years ago)
- Last Synced: 2025-02-01T04:15:26.545Z (9 months ago)
- Language: Python
- Size: 98.1 MB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Oil Riggery
## Objective
Count the number of oil rigs in given areas of the world from satellite imagery.
## Data
Annotated data:
- [An Oil Well Dataset Derived from Satellite-Based Remote Sensing](https://www.mdpi.com/2072-4292/13/6/1132#) called NEPU_OWOD-1.0.
Real-world data for evaluation:
- [Aerial Image Dataset](https://captain-whu.github.io/AID/) a large scale dataset for satellite image classification.
- Create my own dataset. Multi-step process including 1. Using QGIS to create a grid of points over the world. 2. Using Google Earth Engine to download satellite imagery for each point. 3. Using LabelImg to annotate the downloaded images.
## Data exploration
A streamlit data explorer is provided under `/data`. Pre-requisites:
* Add the `NEPU_OWOD-1.0` under the `/data` folder. It can be downloaded following the link above.
Run with:
```bash
poetry run streamlit run oil_riggery/data/Data_explorer.py
```
## Method
TBD