snapshot...

modules/deeplearning/cloud_opd_srcnn_abi.py

@@ -210,7 +210,7 @@ class SRCNN:
         self.test_label_files = None
 
         # self.n_chans = len(data_params_half) + len(data_params_full) + 1
-        self.n_chans = 3
+        self.n_chans = 5
 
         self.X_img = tf.keras.Input(shape=(None, None, self.n_chans))
 
@@ -270,10 +270,11 @@ class SRCNN:
             tmp = normalize(tmp, param, mean_std_dct)
             data_norm.append(tmp)
 
-        tmp = input_label[:, label_idx_i, :, :]
-        tmp = np.where(np.isnan(tmp), 0, tmp)
-        tmp = normalize(tmp, 'refl_0_65um_nom', mean_std_dct)
-        data_norm.append(tmp[:, self.slc_y, self.slc_x])
+        # High res refectance ----------
+        # tmp = input_label[:, label_idx_i, :, :]
+        # tmp = np.where(np.isnan(tmp), 0, tmp)
+        # tmp = normalize(tmp, 'refl_0_65um_nom', mean_std_dct)
+        # data_norm.append(tmp[:, self.slc_y, self.slc_x])
 
         tmp = input_label[:, label_idx_i, :, :]
         tmp = tmp.copy()
@@ -297,16 +298,16 @@ class SRCNN:
         #         tmp = np.where(np.isnan(tmp), 0.0, tmp)
         #     data_norm.append(tmp)
 
-        # for param in sub_fields:
-        #     idx = params.index(param)
-        #     tmp = input_data[:, idx, :, :]
-        #     tmp = upsample_nearest(tmp)
-        #     tmp = tmp[:, self.slc_y, self.slc_x]
-        #     if param != 'refl_substddev_ch01':
-        #         tmp = normalize(tmp, 'refl_0_65um_nom', mean_std_dct)
-        #     else:
-        #         tmp = np.where(np.isnan(tmp), 0, tmp)
-        #     data_norm.append(tmp)
+        for param in sub_fields:
+            idx = params.index(param)
+            tmp = input_data[:, idx, :, :]
+            tmp = upsample_nearest(tmp)
+            tmp = tmp[:, self.slc_y, self.slc_x]
+            if param != 'refl_substddev_ch01':
+                tmp = normalize(tmp, 'refl_0_65um_nom', mean_std_dct)
+            else:
+                tmp = np.where(np.isnan(tmp), 0, tmp)
+            data_norm.append(tmp)
         # ---------------------------------------------------
 
         data = np.stack(data_norm, axis=3)
@@ -454,7 +455,7 @@ class SRCNN:
         self.loss = tf.keras.losses.MeanSquaredError()  # Regression
 
         # decayed_learning_rate = learning_rate * decay_rate ^ (global_step / decay_steps)
-        initial_learning_rate = 0.002
+        initial_learning_rate = 0.001
         decay_rate = 0.95
         steps_per_epoch = int(self.num_data_samples/BATCH_SIZE)  # one epoch
         decay_steps = int(steps_per_epoch) * 4