snapshot..

modules/deeplearning/srcnn_l1b_l2.py

@@ -24,12 +24,12 @@ else:
     NumLogits = NumClasses
 
 BATCH_SIZE = 128
-NUM_EPOCHS = 80
+NUM_EPOCHS = 60
 
 TRACK_MOVING_AVERAGE = False
 EARLY_STOP = True
 
-NOISE_TRAINING = True
+NOISE_TRAINING = False
 NOISE_STDDEV = 0.001
 DO_AUGMENT = True
 
@@ -270,9 +270,9 @@ class SRCNN:
         else:
             if DO_ADD_NOISE:
                 tmp = add_noise(tmp, noise_scale=NOISE_STDDEV)
+                tmp = np.where(tmp < 0.0, 0.0, tmp)
+                tmp = np.where(tmp > 1.0, 1.0, tmp)
             tmp = np.where(np.isnan(tmp), 0, tmp)
-            tmp = np.where(tmp < 0.0, 0.0, tmp)
-            tmp = np.where(tmp > 1.0, 1.0, tmp)
         tmp = resample_2d_linear(x_2, y_2, tmp, t, s)
         data_norm.append(tmp)
         # ---------
@@ -408,7 +408,7 @@ class SRCNN:
         activation = tf.nn.relu
         momentum = 0.99
 
-        num_filters = 128
+        num_filters = 32
 
         input_2d = self.inputs[0]
         print('input: ', input_2d.shape)
@@ -740,7 +740,8 @@ def run_evaluate_static(in_file, out_file, ckpt_dir):
     grd_c = resample_2d_linear_one(x_2, y_2, grd_c, t, s)
     print(grd_a.shape, grd_b.shape, grd_c.shape)
 
-    data = np.stack([grd_a, grd_b, grd_c], axis=2)
+    # data = np.stack([grd_a, grd_b, grd_c], axis=2)
+    data = np.stack([grd_c], axis=2)
     data = np.expand_dims(data, axis=0)
 
     nn = SRCNN()