Replaced deprecated scipy image-resize with PIL.

Author: Hvass-Labs

reinforcement_learning.py

@@ -158,7 +158,7 @@
 import numpy as np
 import tensorflow as tf
 import gym
-import scipy
+import PIL.Image
 import sys
 import os
 import time
@@ -405,6 +405,9 @@ def print_progress(msg):
 # Size of each image in the state.
 state_img_size = np.array([state_height, state_width])
 
+# Size of each image in the state. Reversed order used by PIL.Image.
+state_img_size_reverse = tuple(reversed(state_img_size))
+
 # Number of images in the state.
 state_channels = 2
 
@@ -435,12 +438,19 @@ def _pre_process_image(image):
  """Pre-process a raw image from the game-environment."""
 
  # Convert image to gray-scale.
- img = _rgb_to_grayscale(image)
+ img_gray = _rgb_to_grayscale(image=image)
+
+ # Create PIL-object from numpy array.
+ img = PIL.Image.fromarray(img_gray)
+
+ # Resize the image.
+ img_resized = img.resize(size=state_img_size_reverse,
+ resample=PIL.Image.LINEAR)
 
- # Resize to the desired size using SciPy for convenience.
- img = scipy.misc.imresize(img, size=state_img_size, interp='bicubic')
+ # Convert 8-bit pixel values back to floating-point.
+ img_resized = np.float32(img_resized)
 
- return img
+ return img_resized
 
 
 class MotionTracer: