snapshot...

modules/deeplearning/cnn_cld_frac_mod_res.py

@@ -62,10 +62,11 @@ IMG_DEPTH = 1
 label_param = 'cloud_probability'
 
 params = ['temp_11_0um_nom', 'refl_0_65um_nom', label_param]
+params_i = ['refl_0_65um_nom', label_param]
 data_params_half = ['temp_11_0um_nom']
 data_params_full = ['refl_0_65um_nom']
 
-label_idx = params.index(label_param)
+label_idx = params_i.index(label_param)
 # label_idx = 0
 
 print('data_params_half: ', data_params_half)
@@ -350,7 +351,6 @@ class SRCNN:
             label_s.append(nda)
         input_data = np.concatenate(data_s)
         input_label = np.concatenate(label_s)
-        input_label = input_label[:, 0, :, :]
 
         data_norm = []
         for param in data_params_half:
@@ -366,8 +366,9 @@ class SRCNN:
             data_norm.append(tmp)
 
         for param in data_params_full:
-            idx = params.index(param)
-            tmp = input_data[:, idx, :, :]
+            idx = params_i.index(param)
+            # tmp = input_data[:, idx, :, :]
+            tmp = input_label[:, idx, :, :]
             tmp = tmp.copy()
 
             lo, hi, std, avg = get_min_max_std(tmp)
@@ -381,7 +382,8 @@ class SRCNN:
             data_norm.append(avg[:, 0:66, 0:66])
             # data_norm.append(std[:, 0:66, 0:66])
         # ---------------------------------------------------
-        tmp = input_data[:, label_idx, :, :]
+        # tmp = input_data[:, label_idx, :, :]
+        tmp = input_data[:, 2, :, :]
         tmp = tmp.copy()
         tmp = np.where(np.isnan(tmp), 0, tmp)
         if DO_ESPCN:
@@ -399,7 +401,7 @@ class SRCNN:
         data = data.astype(np.float32)
         # -----------------------------------------------------
         # -----------------------------------------------------
-        label = input_label
+        label = input_label[:, label_idx, :, :]
         label = label.copy()
         label = label[:, y_128, x_128]
         if NumClasses == 5:
@@ -799,10 +801,10 @@ class SRCNN:
         return pred
 
     def run(self, directory, ckpt_dir=None, num_data_samples=50000):
-        train_data_files = glob.glob(directory+'data_train*.npy')
-        valid_data_files = glob.glob(directory+'data_valid*.npy')
-        train_label_files = glob.glob(directory+'label_train*.npy')
-        valid_label_files = glob.glob(directory+'label_valid*.npy')
+        train_data_files = glob.glob(directory+'train_mres_*.npy')
+        valid_data_files = glob.glob(directory+'valid_mres*.npy')
+        train_label_files = glob.glob(directory+'train_ires*.npy')
+        valid_label_files = glob.glob(directory+'valid_ires_*.npy')
         self.setup_pipeline(train_data_files, train_label_files, valid_data_files, valid_label_files, num_data_samples)
 
         # train_data_files = glob.glob(directory+'data_train_*.npy')