HybridSN
This is a Keras reproduction of paper:HybridSN: Exploring 3D-2D CNN Feature Hierarchy for Hyperspectral Image Classification
Test result:
IN dataset
patch size = 25, bands after PCA = 30, train:validation:test = 0.2:0.1:0.7
Test loss:0.049681294709444046
Test acc:98.98954629898071%
Classification result:
precision recall f1-score support
Alfalfa 1.00 0.96 0.98 26
Corn-notill 1.00 0.99 1.00 800
Corn-mintill 0.99 0.97 0.98 465
Corn 0.99 1.00 0.99 133
Grass-pasture 0.95 0.99 0.97 270
Grass-trees 0.98 0.99 0.98 409
Grass-pasture-mowed 1.00 0.80 0.89 15
Hay-windrowed 1.00 1.00 1.00 267
Oats 1.00 0.82 0.90 11
Soybean-notill 0.99 0.99 0.99 545
Soybean-mintill 1.00 0.99 0.99 1375
Soybean-clean 0.97 1.00 0.99 333
Wheat 0.98 0.99 0.99 115
Woods 0.99 1.00 0.99 708
Building-Gras-Tree-Drive 1.00 0.99 1.00 216
Stone-Steel-Towers 1.00 0.98 0.99 52
accuracy 0.99 5740
macro avg 0.99 0.97 0.98 5740
weighted avg 0.99 0.99 0.99 5740Confusion matrix:
Prediction Map:
SA dataset:
patch size = 25, bands after PCA = 15, train:validation:test = 0.2:0.1:0.7
Test loss:0.00021145949722267687
Test acc:99.99340176582336%
Classification result:
precision recall f1-score support
Broc green weeds 1 1.00 1.00 1.00 1125
Broc green weeds 22 1.00 1.00 1.00 2087
Fallow 1.00 1.00 1.00 1107
Fallow rough plow 1.00 1.00 1.00 781
Fallow smooth 1.00 1.00 1.00 1499
Stubble 1.00 1.00 1.00 2217
Celery 1.00 1.00 1.00 2004
Grapes untrained 1.00 1.00 1.00 6312
Soy vineyard develop 1.00 1.00 1.00 3474
Corn sen green weeds 1.00 1.00 1.00 1835
Lettuce romaine 4wk 1.00 1.00 1.00 598
Lettuce romaine 5wk 1.00 1.00 1.00 1079
Lettuce romaine 6wk 1.00 1.00 1.00 513
Lettuce romaine 7wk 1.00 1.00 1.00 599
Vineyard untrained 1.00 1.00 1.00 4071
Vyard verti trellis 1.00 1.00 1.00 1012
accuracy 1.00 30313
macro avg 1.00 1.00 1.00 30313
weighted avg 1.00 1.00 1.00 30313Confusion matrix:
Prediction map:
