Added data normalization. Computation of data mean and standard deviation

DataPreparation/torcs_data_preparation.py

@@ -4,6 +4,7 @@ import csv
 import cv2
 import datetime
 import h5py
+import matplotlib.pyplot as plt
 import mxnet as mx
 import numpy as np
 from PIL import Image
@@ -12,6 +13,9 @@ from sklearn.cross_validation import train_test_split
 import tarfile
 import os
 
+TEST_REC = "torcs_test.rec"
+TRAIN_REC = "torcs_train.rec"
+
 ARCHIVE = False
 CHUNK_SIZE = 10000
 LEVELDB_PATH = "/media/sveta/4991e634-dd81-4cb9-bf46-2fa9c7159263/TORCS_Training_1F"
@@ -19,34 +23,63 @@ HDF5_PATH = "/media/sveta/4991e634-dd81-4cb9-bf46-2fa9c7159263/TORCS_HDF5_3/"
 RAW_PATH = "/media/sveta/4991e634-dd81-4cb9-bf46-2fa9c7159263/TORCS_raw/"
 EXAMPLES_PATH = "/media/sveta/4991e634-dd81-4cb9-bf46-2fa9c7159263/TORCS_examples/"
 
+TRAIN_MEAN = [99.39394537, 110.60877108, 117.86127587]
+TRAIN_STD = [42.04910545, 49.47874084, 62.61726178]
 
 def main():
     start_date = datetime.datetime.now()
 
     # leveldb_to_rec(start_date)
-    read_from_recordio()
+    # read_from_recordio()
     # compute_train_mean()
+    test_normalization()
 
 
 def compute_train_mean():
-    record = mx.recordio.MXRecordIO(RAW_PATH + "torcs_train.rec", "r")
-    img_stats = []
-    for i in range(50):
+    record = mx.recordio.MXRecordIO(RAW_PATH + TRAIN_REC, "r")
+    all_means = list()
+    all_std = list()
+    for i in range(387851):
+        if i % 1000 == 0:
+            print(i)
         item = record.read()
         header, img_as_array = mx.recordio.unpack_img(item)
         # img is RGB of shape (210, 280, 3)
-        # img = Image.fromarray(img_as_array)
-        # img.show()
         mean, std = cv2.meanStdDev(img_as_array)
-        mean_as_img = Image.fromarray(mean)
-        mean_as_img.show()
-        img_stats.append(np.array([mean[::-1] / 255, std[::-1] / 255]))
-    img_stats = np.mean(img_stats, axis=0)
-    print(img_stats)
-
+        all_means.append(mean)
+        all_std.append(std)
+        # img_stats.append(np.array([mean[::-1] / 255, std[::-1] / 255]))
+    mean = np.mean(all_means, axis=0)
+    print("MEAN")
+    print(mean)
+    std = np.mean(all_std, axis=0)
+    print("STD")
+    print(std)
+
+
+def test_normalization():
+    width = 280
+    height = 210
+    data_mean = np.asarray([[[a] * width] * height for a in TRAIN_MEAN]) # (3, 280, 210)
+    data_std = np.asarray([[[a] * width] * height for a in TRAIN_STD])
+    record = mx.recordio.MXRecordIO(RAW_PATH + TRAIN_REC, "r")
+    item = record.read()
+    header, img_as_array = mx.recordio.unpack_img(item) # (210, 280,3)
+    img = Image.fromarray(img_as_array)
+    img.show()
+
+    normalized_img = (img_as_array - np.transpose(data_mean, (1, 2, 0)))/np.transpose(data_std, (1, 2, 0))
+    plt.matshow(normalized_img[:,:,0])
+    plt.show()
+
+    plt.matshow(normalized_img[:,:,1])
+    plt.show()
+
+    plt.matshow(normalized_img[:,:,2])
+    plt.show()
 
 def read_from_recordio():
-    record = mx.recordio.MXRecordIO(RAW_PATH + "torcs_train.rec", "r")
+    record = mx.recordio.MXRecordIO(RAW_PATH + TRAIN_REC, "r")
     for i in range(50):
         item = record.read()
         header, img = mx.recordio.unpack_img(item)
@@ -60,8 +93,8 @@ def read_from_recordio():
 
 
 def leveldb_to_rec(start_date):
-    train_record = mx.recordio.MXRecordIO(RAW_PATH + "torcs_train.rec", "w")
-    test_record = mx.recordio.MXRecordIO(RAW_PATH + "torcs_test.rec", "w")
+    train_record = mx.recordio.MXRecordIO(RAW_PATH + TRAIN_REC, "w")
+    test_record = mx.recordio.MXRecordIO(RAW_PATH + TEST_REC, "w")
     keys = range(1, 484815)
 
     train_keys, test_keys = train_test_split(keys,test_size=0.2)

GeneratedDpnetTrainingCode/CNNCreator_dpnet.py

@@ -84,7 +84,7 @@ class CNNCreator_dpnet:
     def load_h5_files(self):
         train_h5 = None
         test_h5 = None
-        train_path = self._data_dir_ + "train_2.h5"
+        train_path = self._data_dir_ + "train.h5"
         test_path = self._data_dir_ + "test.h5"
         if os.path.isfile(train_path):
             train_h5 = h5py.File(train_path, 'r')
@@ -113,7 +113,7 @@ class CNNCreator_dpnet:
               optimizer_params=(('learning_rate', 0.001),),
               load_checkpoint=True,
               context='gpu',
-              checkpoint_period=1,
+              checkpoint_period=5,
               normalize=True):
         if context == 'gpu':
             mx_context = mx.gpu()
@@ -161,8 +161,8 @@ class CNNCreator_dpnet:
 
         self.module.fit(
             train_data=train_iter,
-            eval_metric='mse',
             eval_data=test_iter,
+            eval_metric='mse',
             optimizer=optimizer,
             optimizer_params=optimizer_params,
             batch_end_callback=mx.callback.Speedometer(batch_size),
@@ -334,8 +334,3 @@ class CNNCreator_dpnet:
                                     data_names=self._input_names_,
                                     label_names=self._output_names_,
                                     context=context)
-        # print("start viz")
-        # graph = mx.viz.plot_network(predictions, shape={'image':(0,3,210,280), 'predictions_label':(0,14)}, node_attrs={"shape":'rect',"fixedsize":'false'})
-        # # graph.format = 'png'
-        # graph.render('graph')
-

GeneratedDpnetTrainingCode/CNNTrainer_Dpnet.py

@@ -5,18 +5,23 @@ import CNNCreator_dpnet
 if __name__ == "__main__":
     logging.basicConfig(level=logging.DEBUG)
     logger = logging.getLogger()
-    handler = logging.FileHandler("train.log","w+", encoding=None, delay="true")
+    handler = logging.FileHandler("train.log","w", encoding=None, delay="true")
     logger.addHandler(handler)
 
     dpnet = CNNCreator_dpnet.CNNCreator_dpnet()
     dpnet.train(
         batch_size = 64,
-        num_epoch = 10,
+        num_epoch = 100,
         context = 'cpu',
-        normalize = False,
+        normalize = True,
         optimizer = 'sgd',
         optimizer_params = {
             'weight_decay': 5.0E-4,
-            'learning_rate': 0.01}
+            'learning_rate_minimum': 0.01,
+            'learning_rate_decay': 0.9,
+            'step_size': 8000}
+
+
+
     )
 

GeneratedDpnetTrainingCode/custom_functions.py

 import mxnet as mx
+import numpy as np
+
+TRAIN_MEAN = [99.39394537, 110.60877108, 117.86127587]
+TRAIN_STD = [42.04910545, 49.47874084, 62.61726178]
 
 
 def load_data_rec(self, batch_size):
+    width = 280
+    height = 210
 
     train_iter = mx.image.ImageIter(
         path_imgrec=self._data_dir_ + "torcs_train.rec",
-        data_shape=(3, 210, 280),  # (channels, height, width)
+        data_shape=(3, height, width),  # (channels, height, width)
         batch_size=batch_size,
         label_width=14,
         data_name='image',
@@ -13,14 +19,14 @@ def load_data_rec(self, batch_size):
     )
     test_iter = mx.image.ImageIter(
         path_imgrec=self._data_dir_ + "torcs_test.rec",
-        data_shape=(3, 210, 280),  # (channels, height, width)
+        data_shape=(3, height, width),  # (channels, height, width)
         batch_size=batch_size,
         label_width=14,
         data_name='image',
         label_name='predictions_label'
     )
-    data_mean = None
-    data_std = None
 
+    data_mean = np.asarray([[[a] * width] * height for a in TRAIN_MEAN])
+    data_std = np.asarray([[[a] * width] * height for a in TRAIN_STD])
 
     return train_iter, test_iter, data_mean, data_std

GeneratedDpnetTrainingCode/requirements-cpu.txt

+mxnet
+h5py
\ No newline at end of file

GeneratedDpnetTrainingCode/requirements_gpu.txt

+mxnet-cu75
+h5py
\ No newline at end of file