0

I want to do crop classification using deep learning methods. I have collected the dataset of crop types in Ghana. My dataset contains tiff images and json files. The tiff file is not georeferenced. The json file looks like this:

{"type":"Feature","geometry":{"type":"Polygon","coordinates":[[[-2,8],[1,8],[1,11],[-2,11],[-2,8]]]},"properties":{"anon:size":3,"anon:warning":"WARNING: The item geometries in this collection have been anonymized such that they exist somewhere within the bounding box listed within that item's geometry.","datetime":"2017-01-02T00:00:00","id":"su_african_crops_ghana_source_planet_000000_2017_01_02"}}
]}```

How to make the tiff image georeferenced using this json file for my deep learning project. I want to do pixel-wise classification. Can anyone help to make the tiff image georeferenced? Is there any way to do it in python? 
Thank you.
Rameen Rahman
  • 11
  • 1
  • 2
  • 4

1 Answers1

0

You will need to follow abit of steps:

Open GeoJson with list of areas to crop
- Iterate over the the list of area
-- Find the raster that contains the crop area  
-- Create a new raster dataset using metadata from raster you found
-- Get the bounding box from an area to crop from geojson 
-- Read the 3 color bands using the bounding box into numpy array
-- Write the numpy array to the newly created raster
Sheece Gardazi
  • 480
  • 6
  • 14