Sat2samples

A collection of tools to easliy acquire and crop large-scale satellite imagery into samples for machine/deep learning.

Attention: Make sure you have access to Google Earth Engine API.

1. Description

This is an easy-to-use project for acquiring and cutting large-scale satellite imagery into samples for machine/deep learning. This project is mainly based on geemap, gdal, colab and QGIS desktop.

2. Usage

Step 1: Prepare your region of interest (ROI) shapefiles. If your ROI is located in China, please try this link: Get shapefile of ROI.
Step 2: Use QGIS desktop to create the grid of your ROI. If you don't familar with this processing, please explore more in this link: QGIS operation.
Step 3: Upload the shapefile of the grids followed by last step to your GEE assets.
Step 4: Use the notebooks in this project to download Sentinel 1/2 satelite images with preprocessing.
Step 5: Use the .py files in this project to crop the satelite images to samples.

3. Details

Sentinel-1.ipynb: use to download the Sentinel-1_GRD satelite imagery of your ROI.(VV,VH,UInt8)
Sentinel-2.ipynb: use to download the Sentinel-2 satelite imagery of your ROI.
S2_process.py: use to process Sentinel-2 images including float32()toUInt16(),Compress.
dataset_crop/crop_v2.py: use to crop the satelite images to samples such as 512*512.

4. Contact

If you encounter any problem in using the Sat2samples or have any feedback, please contact:

E-mail: 2309162135@geemail.one

5. Reference