Identifying illegal logging in drone imagery using Google Earth Engine
A quick demonstration of on-the-fly land-cover classification of drone-derived orthomosaics using Google Earth Engine for the purpose of delineating illegally logged areas.
A quick demonstration of on-the-fly land-cover classification of drone-derived orthomosaics using Google Earth Engine for the purpose of delineating illegally logged areas.
I’m pretty impressed that it can do supervised classification!
Here’s the steps you need to go through to do this yourself
1. Sign up for Google Earth Engine: https://earthengine.google.com/
2. Go to https://code.earthengine.google.com/
3. In top left panel click ASSETS and upload your ORTHO photo here – this is your “asset”. Note the name
4. Go to https://explorer.earthengine.google.com/#workspace
5. Search for users/[your username]/[name of asset] omitting the square brackets.
6. Your asset should appear in the list so select it and SAVE then zoom to the map
Then, following the video:
7. Click ADD DATA and select HAND DRAWN POINTS AND POLYGONS
8. Click ADD CLASS and label this (e.g. water, soil) – and start marking up your map with either POINTS or POLYGONS selected from tollbox top left of map.
9. Add more classes as required
10. Adjust colour of classes as required
11. Click ANALYSIS and select TRAIN CLASSIFIER
12. Choose your algorithm you want and set the resolution for the output – e.g. 10m
13. Click TRAIN CLASSIFIER AND DISPLAY RESULTS
14. Click on the model that has been added to the DATA list at top left to review the validity table for the model
Here’s the steps you need to go through to do this yourself
1. Sign up for Google Earth Engine: https://earthengine.google.com/
2. Go to https://code.earthengine.google.com/
3. In top left panel click ASSETS and upload your ORTHO photo here – this is your “asset”. Note the name
4. Go to https://explorer.earthengine.google.com/#workspace
5. Search for users/[your username]/[name of asset] omitting the square brackets.
6. Your asset should appear in the list so select it and SAVE then zoom to the map
Then, following the video:
7. Click ADD DATA and select HAND DRAWN POINTS AND POLYGONS
8. Click ADD CLASS and label this (e.g. water, soil) – and start marking up your map with either POINTS or POLYGONS selected from tollbox top left of map.
9. Add more classes as required
10. Adjust colour of classes as required
11. Click ANALYSIS and select TRAIN CLASSIFIER
12. Choose your algorithm you want and set the resolution for the output – e.g. 10m
13. Click TRAIN CLASSIFIER AND DISPLAY RESULTS
14. Click on the model that has been added to the DATA list at top left to review the validity table for the model
Thanks for bringing this to people’s attention! Don’t overlook the new ee.Clusterer toolset either… basically a package of the Weka machine learning suite. Needs API access.
Thank you Adam, that looks like a very interesting tool. If you have some sample code on an example satellite or ortho that would be interesting to see.
Can Anyone kindly provide me more info in this regard and if any team can help me tackling illegal logging in my region