work on do_evaluate...

modules/deeplearning/espcn.py

 import glob
 import tensorflow as tf
-from util.setup import logdir, modeldir, cachepath, now, ancillary_path, home_dir
-from util.util import EarlyStop, normalize, scale, make_for_full_domain_predict
+from util.setup import logdir, modeldir, cachepath, now, ancillary_path
+from util.util import EarlyStop, normalize, denormalize
 import os, datetime
 import numpy as np
 import pickle
@@ -740,7 +740,7 @@ class ESPCN:
 
         self.test_preds = preds
 
-    def do_evaluate(self, nda_lr, ckpt_dir):
+    def do_evaluate(self, nda_lr, param, ckpt_dir):
 
         ckpt = tf.train.Checkpoint(step=tf.Variable(1), model=self.model)
         ckpt_manager = tf.train.CheckpointManager(ckpt, ckpt_dir, max_to_keep=3)
@@ -756,6 +756,8 @@ class ESPCN:
         pred = self.model([data])
         self.test_probs = pred
 
+        return denormalize(pred, param, mean_std_dct[param])
+
     def run(self, directory):
         train_data_files = glob.glob(directory+'data_train*.npy')
         valid_data_files = glob.glob(directory+'data_valid*.npy')