search

DynaSlum / SatelliteImaging

The software for WP1: SatelliteImaging
Apache License 2.0
3 stars 1 forks source link

Split the image datastores into train, validation and test subsets #12

Closed elboyran closed 7 years ago

elboyran commented 7 years ago

Generate train, validation and test sub-image datastores for the 5 image datastores and save them as MAT files:

elboyran commented 7 years ago

`Splitting image data store # 1 out of 5 Training datastore: Label Count

BuiltUp       1075 
Mixed          136 
NonBuiltUp     700 
Slum            63

Testing datastore: Label Count

BuiltUp       230  
Mixed          29  
NonBuiltUp    150  
Slum           14

Validaiton datastore: Label Count

BuiltUp       231  
Mixed          29  
NonBuiltUp    150  
Slum           13

Splitting image data store # 2 out of 5 Training datastore: Label Count

BuiltUp       1681 
Mixed          176 
NonBuiltUp    1126 
Slum           115

Testing datastore: Label Count

BuiltUp       360  
Mixed          38  
NonBuiltUp    241  
Slum           25

Validaiton datastore: Label Count

BuiltUp       360  
Mixed          38  
NonBuiltUp    241  
Slum           24

Splitting image data store # 3 out of 5 Training datastore: Label Count

BuiltUp       2974 
Mixed          267 
NonBuiltUp    2086 
Slum           258

Testing datastore: Label Count

BuiltUp       637  
Mixed          57  
NonBuiltUp    447  
Slum           55

Validaiton datastore: Label Count

BuiltUp       637  
Mixed          58  
NonBuiltUp    447  
Slum           56

Splitting image data store # 4 out of 5 Training datastore: Label Count

BuiltUp       6702 
Mixed          373 
NonBuiltUp    4834 
Slum           753

Testing datastore: Label Count

BuiltUp       1436 
Mixed           80 
NonBuiltUp    1036 
Slum           161

Validaiton datastore: Label Count

BuiltUp       1436 
Mixed           80 
NonBuiltUp    1036 
Slum           161

Splitting image data store # 5 out of 5 Training datastore: Label Count

BuiltUp       27875
Mixed           800
NonBuiltUp    20798
Slum           3816

Testing datastore: Label Count

BuiltUp       5973 
Mixed          171 
NonBuiltUp    4457 
Slum           818

Validaiton datastore: Label Count

BuiltUp       5974 
Mixed          172 
NonBuiltUp    4456 
Slum           818

DONE.`

elboyran commented 7 years ago

Split to Training and Test data only. The validation subset is created using the classificationLearner app's validation scheme!

elboyran commented 7 years ago

Splitting image data store # 1 out of 5 Training datastore: Label Count

BuiltUp       1075 
Mixed          136 
NonBuiltUp     700 
Slum            63

Testing datastore: Label Count

BuiltUp       461  
Mixed          58  
NonBuiltUp    300  
Slum           27

Splitting image data store # 2 out of 5 Training datastore: Label Count

BuiltUp       1681 
Mixed          176 
NonBuiltUp    1126 
Slum           115

Testing datastore: Label Count

BuiltUp       720  
Mixed          76  
NonBuiltUp    482  
Slum           49

Splitting image data store # 3 out of 5 Training datastore: Label Count

BuiltUp       2974 
Mixed          267 
NonBuiltUp    2086 
Slum           258

Testing datastore: Label Count

BuiltUp       1274 
Mixed          115 
NonBuiltUp     894 
Slum           111

Splitting image data store # 4 out of 5 Training datastore: Label Count

BuiltUp       6702 
Mixed          373 
NonBuiltUp    4834 
Slum           753

Testing datastore: Label Count

BuiltUp       2872 
Mixed          160 
NonBuiltUp    2072 
Slum           322

Splitting image data store # 5 out of 5 Training datastore: Label Count

BuiltUp       27875
Mixed           800
NonBuiltUp    20798
Slum           3816

Testing datastore: Label Count

BuiltUp       11947
Mixed           343
NonBuiltUp     8913
Slum           1636

DONE.