svm_classifier.py

import ee 
from ee_plugin import Map 

# Input imagery is a cloud-free Landsat 8 composite.
l8 = ee.ImageCollection('LANDSAT/LC08/C01/T1')

image = ee.Algorithms.Landsat.simpleComposite(**{
  'collection': l8.filterDate('2018-01-01', '2018-12-31'),
  'asFloat': True
})

# Use these bands for prediction.
bands = ['B2', 'B3', 'B4', 'B5', 'B6', 'B7', 'B10', 'B11']

# Manually created polygons.
forest1 = ee.Geometry.Rectangle(-63.0187, -9.3958, -62.9793, -9.3443)
forest2 = ee.Geometry.Rectangle(-62.8145, -9.206, -62.7688, -9.1735)
nonForest1 = ee.Geometry.Rectangle(-62.8161, -9.5001, -62.7921, -9.4486)
nonForest2 = ee.Geometry.Rectangle(-62.6788, -9.044, -62.6459, -8.9986)

# Make a FeatureCollection from the hand-made geometries.
polygons = ee.FeatureCollection([
  ee.Feature(nonForest1, {'class': 0}),
  ee.Feature(nonForest2, {'class': 0}),
  ee.Feature(forest1, {'class': 1}),
  ee.Feature(forest2, {'class': 1}),
])

# Get the values for all pixels in each polygon in the training.
training = image.sampleRegions(**{
  # Get the sample from the polygons FeatureCollection.
  'collection': polygons,
  # Keep this list of properties from the polygons.
  'properties': ['class'],
  # Set the scale to get Landsat pixels in the polygons.
  'scale': 30
})

# Create an SVM classifier with custom parameters.
classifier = ee.Classifier.libsvm(**{
  'kernelType': 'RBF',
  'gamma': 0.5,
  'cost': 10
})

# Train the classifier.
trained = classifier.train(training, 'class', bands)

# Classify the image.
classified = image.classify(trained)

# Display the classification result and the input image.
Map.setCenter(-62.836, -9.2399, 9)
Map.addLayer(image, {'bands': ['B4', 'B3', 'B2'], 'max': 0.5, 'gamma': 2})
Map.addLayer(polygons, {}, 'training polygons')
Map.addLayer(classified,
             {'min': 0, 'max': 1, 'palette': ['red', 'green']},
             'deforestation')