Added a flag whether to compute the confusion matrix

main/config.py

@@ -21,12 +21,12 @@ import torch.optim as optim
 wd = '/mnt/CEPH_PROJECTS/cci_snow/dfrisinghelli/'
 
 # define which dataset to train on
-dataset_name = 'Sparcs'
-# dataset_name= 'Cloud95'
+# dataset_name = 'Sparcs'
+dataset_name= 'Cloud95'
 
 # path to the dataset
-dataset_path = os.path.join(wd, '_Datasets/Sparcs')
-# dataset_path = os.path.join(wd, '_Datasets/Cloud95/Training')
+# dataset_path = os.path.join(wd, '_Datasets/Sparcs')
+dataset_path = os.path.join(wd, '_Datasets/Cloud95/Training')
 
 # the csv file containing the names of the informative patches of the
 # Cloud95 dataset
@@ -39,7 +39,7 @@ bands = ['red', 'green', 'blue', 'nir']
 
 # define the size of the network input
 # if None, the size will default to the size of a scene
-tile_size = 125
+tile_size = 192
 
 # -----------------------------------------------------------------------------
 # -----------------------------------------------------------------------------
@@ -76,7 +76,7 @@ ttratio = 1
 
 # (ttratio * tvratio) * 100 % will be used as the training dataset
 # (1 - ttratio * tvratio) * 100 % will be used as the validation dataset
-tvratio = 0.8
+tvratio = 0.2
 
 # define the batch size
 # determines how many samples of the dataset are processed until the weights
@@ -88,14 +88,14 @@ checkpoint = False
 
 # whether to early stop training if the accuracy (loss) on the validation set
 # does not increase (decrease) more than delta over patience epochs
-early_stop = True
+early_stop = False
 mode = 'max'
-delta = 0.005
+delta = 0
 patience = 10
 
 # define the number of epochs: the number of maximum iterations over the whole
 # training dataset
-epochs = 200
+epochs = 5
 
 # define the number of threads
 nthreads = os.cpu_count()
@@ -117,6 +117,9 @@ state_path = os.path.join(os.getcwd(), '_models/')
 # ------------------------- Plotting configuration ----------------------------
 # -----------------------------------------------------------------------------
 
+# whether to compute and plot confusion matrix for the entire validation set
+plot_cm = False
+
 # whether to save plots of (input, ground truth, prediction) of the validation
 # dataset to disk
 # output path is: current_working_directory/_samples/

main/eval.py

@@ -12,23 +12,25 @@ import matplotlib.pyplot as plt
 sys.path.append('..')
 
 # local modules
-from main.config import state_path, plot_samples, nsamples, plot_bands, seed
+from main.config import (state_path, plot_samples, nsamples, plot_bands, seed,
+                         plot_cm)
 from main.init import state_file, trainer
 
 if __name__ == '__main__':
 
-    # predict each batch in the validation set
-    cm, accuracy, loss = trainer.predict(state_path, state_file,
-                                         confusion=True)
+    if plot_cm:
+        # predict each batch in the validation set
+        cm, accuracy, loss = trainer.predict(state_path, state_file,
+                                             confusion=True)
 
-    # # calculate overal accuracy
-    acc = (cm.diag().sum() / cm.sum()).numpy().item()
-    print('After training for {:d} epochs, we achieved an overall accuracy of '
-          '{:.2f}%  on the validation set!'.format(trainer.model.epoch,
-                                                   acc * 100))
+        # calculate overal accuracy
+        acc = (cm.diag().sum() / cm.sum()).numpy().item()
+        print('After training for {:d} epochs, we achieved an overall '
+              'accuracy of {:.2f}%  on the validation set!'
+              .format(trainer.model.epoch, acc * 100))
 
-    # # plot confusion matrix
-    trainer.dataset.plot_confusion_matrix(cm, state=state_file)
+        # plot confusion matrix
+        trainer.dataset.plot_confusion_matrix(cm, state=state_file)
 
     # plot loss and accuracy
     trainer.dataset.plot_loss(
@@ -37,6 +39,10 @@ if __name__ == '__main__':
     # whether to plot the samples of the validation dataset
     if plot_samples:
 
+        # load pretrained model
+        state = trainer.model.load(trainer.optimizer, state_file, state_path)
+        trainer.model.eval()
+
         # base filename for each sample
         fname = state_file.split('.pt')[0]
 
@@ -68,4 +74,4 @@ if __name__ == '__main__':
                                                   bands=plot_bands,
                                                   state=sname,
                                                   stretch=True,
-						  alpha=5)
+                                                  alpha=5)