Add salinas dataset

datasets.py

@@ -26,6 +26,12 @@
             'img': 'Pavia.mat',
             'gt': 'Pavia_gt.mat'
             },
+        'Salinas': {
+            'urls': ['http://www.ehu.eus/ccwintco/uploads/a/a3/Salinas_corrected.mat',
+                     'http://www.ehu.eus/ccwintco/uploads/f/fa/Salinas_gt.mat'],
+            'img': 'Salinas_corrected.mat',
+            'gt': 'Salinas_gt.mat'
+            },
         'PaviaU': {
             'urls': ['http://www.ehu.eus/ccwintco/uploads/e/ee/PaviaU.mat',
                      'http://www.ehu.eus/ccwintco/uploads/5/50/PaviaU_gt.mat'],
@@ -138,6 +144,20 @@ def get_dataset(dataset_name, target_folder="./", datasets=DATASETS_CONFIG):
 
         ignored_labels = [0]
 
+
+    elif dataset_name == 'Salinas':
+        img = open_file(folder + 'Salinas.mat')['Salinas_corrected']
+
+        rgb_bands = (43, 21, 11)  # AVIRIS sensor
+
+        gt = open_file(folder + 'Salinas_gt.mat')['Salinas_gt']
+
+        label_values = ['Undefined','Brocoli_green_weeds_1', 'Brocoli_green_weeds_2','Fallow','Fallow_rough_plow','Fallow_smooth','Stubble','Celery',
+                       'Grapes_untrained','Soil_vinyard_develop','Corn_senesced_green_weeds','Lettuce_romaine_4wk','Lettuce_romaine_5wk','Lettuce_romaine_6wk',
+                       'Lettuce_romaine_7wk','Vinyard_untrained','Vinyard_vertical_trellis']
+
+        ignored_labels = [0]    
+
     elif dataset_name == 'IndianPines':
         # Load the image
         img = open_file(folder + 'Indian_pines_corrected.mat')

inference.py

@@ -1,7 +1,7 @@
 # Python 2/3 compatiblity
 from __future__ import print_function
 from __future__ import division
-
+import joblib
 import os
 from utils import convert_to_color_, convert_from_color_, get_device
 from datasets import open_file
@@ -91,7 +91,6 @@ def convert_from_color(x):
     return convert_from_color_(x, palette=invert_palette)
 
 if MODEL in ['SVM', 'SVM_grid', 'SGD', 'nearest']:
-    from sklearn.externals import joblib
     model = joblib.load(CHECKPOINT)
     w, h = img.shape[:2]
     X = img.reshape((w*h, N_BANDS))

requirements.txt

@@ -9,3 +9,4 @@ scikit-image>=0.13.1
 torch>=0.4.0
 matplotlib>=2.0.2
 torchsummary>=1.5
+joblib==0.14.1