Added validation loss output

driving_data.py

@@ -5,7 +5,8 @@
 ys = []
 
 #points to the end of the last batch
-batch_pointer = 0
+train_batch_pointer = 0
+val_batch_pointer = 0
 
 #read data.txt
 with open("driving_dataset/data.txt") as f:
@@ -24,12 +25,31 @@
 random.shuffle(c)
 xs, ys = zip(*c)
 
-def LoadBatch(batch_size):
-    global batch_pointer
+train_xs = xs[:int(len(xs) * 0.8)]
+train_ys = ys[:int(len(xs) * 0.8)]
+
+val_xs = xs[-int(len(xs) * 0.2):]
+val_ys = ys[-int(len(xs) * 0.2):]
+
+num_train_images = len(train_xs)
+num_val_images = len(val_xs)
+
+def LoadTrainBatch(batch_size):
+    global train_batch_pointer
+    x_out = []
+    y_out = []
+    for i in range(0, batch_size):
+        x_out.append(scipy.misc.imresize(scipy.misc.imread(train_xs[(train_batch_pointer + i) % num_train_images])[-150:], [66, 200]) / 255.0)
+        y_out.append([train_ys[(train_batch_pointer + i) % num_train_images]])
+    train_batch_pointer += batch_size
+    return x_out, y_out
+
+def LoadValBatch(batch_size):
+    global val_batch_pointer
     x_out = []
     y_out = []
     for i in range(0, batch_size):
-        x_out.append(scipy.misc.imresize(scipy.misc.imread(xs[(batch_pointer + i) % num_images])[-150:], [66, 200]) / 255.0)
-        y_out.append([ys[(batch_pointer + i) % num_images]])
-    batch_pointer += batch_size
+        x_out.append(scipy.misc.imresize(scipy.misc.imread(val_xs[(val_batch_pointer + i) % num_val_images])[-150:], [66, 200]) / 255.0)
+        y_out.append([val_ys[(val_batch_pointer + i) % num_val_images]])
+    val_batch_pointer += batch_size
     return x_out, y_out

train.py

@@ -15,10 +15,11 @@
 
 #train over the dataset about 30 times
 for i in range(int(driving_data.num_images * 0.3)):
-  xs, ys = driving_data.LoadBatch(100)
+  xs, ys = driving_data.LoadTrainBatch(100)
   train_step.run(feed_dict={model.x: xs, model.y_: ys, model.keep_prob: 0.8})
   if i % 10 == 0:
-    print("step %d, train loss %g"%(i, loss.eval(feed_dict={
+    xs, ys = driving_data.LoadValBatch(100)
+    print("step %d, val loss %g"%(i, loss.eval(feed_dict={
          model.x:xs, model.y_: ys, model.keep_prob: 1.0})))
   if i % 100 == 0:
     if not os.path.exists(LOGDIR):