[Angelica] Extract feature and data modules/methods into separate dir…

…ectories

.gitignore

@@ -6,4 +6,5 @@ pip-selfcheck.json
 share/
 dist/
 Riot_python.egg-info
-*.pyc
+*.pyc
+build/

data/dataProcessor.py

@@ -0,0 +1,53 @@
+import os
+from feature import Feature
+import numpy as np
+
+
+class DataProcessor:
+    """
+    Class containing all necessary data preprocessing methods.
+    """
+
+    def get_image_feature_array(self, root_directory, vector=True):
+        """
+        Extracts features vectors of all images found in root_directory.
+        :param root_directory: location of image data
+        :param vector: if true returns features as vectors, otherwise as 2D arrays
+        :return: numpy array of extracted feature vectors
+        """
+        feature_index = 0 if vector else 1
+        feature = Feature()
+        features = list()
+        for sub_directory in os.listdir(root_directory):
+            if not sub_directory.startswith('.'):
+                sub_directory_path = root_directory + '/' + sub_directory
+                for image_file in os.listdir(sub_directory_path):
+                    if not image_file.startswith('.'):
+                        image_file_path = sub_directory_path + '/' + image_file
+                        features.append([[feature.extract_hog_feature_vector(image_file_path)[feature_index]]])
+
+        return np.array(features)
+
+    def get_time_series_image_feature_array(self, root_directory, vector=True):
+        """
+        Extracts features vectors of images found in root_directory and groups them
+        by time_series batch. Subdirectories of root_directory must contain a single
+        time series batch.
+        :param root_directory: location of image data
+        :param vector: if true returns features as vectors, otherwise as 2D arrays
+        :return: numpy array of arrays which contain time series batch features
+        """
+        feature_index = 0 if vector else 1
+        feature = Feature()
+        features = list()
+        for sub_directory in os.listdir(root_directory):
+            if not sub_directory.startswith('.'):
+                sub_directory_path = root_directory + '/' + sub_directory
+                for image_file in os.listdir(sub_directory_path):
+                    feature_batch = list()
+                    if not image_file.startswith('.'):
+                        image_file_path = sub_directory_path + '/' + image_file
+                        feature_batch.append(feature.extract_hog_feature_vector(image_file_path)[feature_index])
+                features.append(feature_batch)
+
+        return np.array(features)

data/fer2013/README

@@ -0,0 +1,10 @@
+If you use this dataset in your research work, please cite
+
+"Challenges in Representation Learning: A report on three machine learning
+contests." I Goodfellow, D Erhan, PL Carrier, A Courville, M Mirza, B
+Hamner, W Cukierski, Y Tang, DH Lee, Y Zhou, C Ramaiah, F Feng, R Li,
+X Wang, D Athanasakis, J Shawe-Taylor, M Milakov, J Park, R Ionescu,
+M Popescu, C Grozea, J Bergstra, J Xie, L Romaszko, B Xu, Z Chuang, and
+Y. Bengio. arXiv 2013.
+
+See fer2013.bib for a bibtex entry.

data/fer2013/fer2013.bib

@@ -0,0 +1,19 @@
+
+@MISC{Goodfeli-et-al-2013,
+       author = {Goodfellow, Ian and Erhan, Dumitru and Carrier, Pierre-Luc and Courville, Aaron and Mirza, Mehdi and Hamner, Ben and Cukierski, Will and Tang, Yichuan and Thaler, David and Lee, Dong-Hyun and Zhou, Yingbo and Ramaiah, Chetan and Feng, Fangxiang and Li, Ruifan and Wang, Xiaojie and Athanasakis, Dimitris and Shawe-Taylor, John and Milakov, Maxim and Park, John and Ionescu, Radu and Popescu, Marius and Grozea, Cristian and Bergstra, James and Xie, Jingjing and Romaszko, Lukasz and Xu, Bing and Chuang, Zhang and Bengio, Yoshua},
+     keywords = {competition, dataset, representation learning},
+        title = {Challenges in Representation Learning: A report on three machine learning contests},
+         year = {2013},
+  institution = {Unicer},
+          url = {http://arxiv.org/abs/1307.0414},
+     abstract = {The ICML 2013 Workshop on Challenges in Representation
+Learning focused on three challenges: the black box learning challenge,
+the facial expression recognition challenge, and the multimodal learn-
+ing challenge. We describe the datasets created for these challenges and
+summarize the results of the competitions. We provide suggestions for or-
+ganizers of future challenges and some comments on what kind of knowl-
+edge can be gained from machine learning competitions.
+
+http://deeplearning.net/icml2013-workshop-competition}
+}
+

data/temp.py

@@ -0,0 +1,10 @@
+import sys
+sys.path.append('../feature')
+from dataProcessor import DataProcessor
+
+
+d = DataProcessor()
+root_directory = "../data/cohn_kanade_images"
+feature_images = d.get_time_series_image_feature_array(root_directory)
+
+print(feature_images.shape)

experiments/conv_lstm.py

@@ -1,3 +1,5 @@
+import sys
+sys.path.append('../feature')
 from keras.layers import ConvLSTM2D, Dense, Flatten
 from keras.models import Sequential
 from keras.callbacks import ReduceLROnPlateau, EarlyStopping

experiments/data.py

@@ -8,7 +8,7 @@
 
 def get_image_feature_vector_array():
     feature = Feature()
-    root_directory = "../images"
+    root_directory = "../data/cohn_kanade_images"
     features = list()
     for subfile in os.listdir(root_directory):
         if "DS_Store" in subfile:   continue
@@ -21,7 +21,7 @@ def get_image_feature_vector_array():
 
 def get_image_feature_image_array():
     feature = Feature()
-    root_directory = "../images"
+    root_directory = "../data/cohn_kanade_images"
     features = list()
     for subfile in os.listdir(root_directory):
         if "DS_Store" in subfile:   continue
@@ -32,6 +32,39 @@ def get_image_feature_image_array():
 
     return np.array(features)
 
+def get_raw_training_labels():
+    # Uses 20 photo series from the Cohn-Kanade dataset
+    # hand labeled by AP
+    # arousal(least, most), valence(negative, positive), power, anticipation
+    raw_training_labels = {1: [10, [.6, .4, .7, .6], [.9, .1, .8, .9]],
+                           2: [9, [.2, .5, .6, .1], [.3, .4, .5, .2]],
+                           3: [10, [.8, .9, .2, .9], [.99, .99, .1, .99]],
+                           4: [10, [.2, .4, .4, .5], [.8, .2, .7, .6]],
+                           5: [8, [.2, .4, .2, .1], [.5, .5, .5, .5]],
+                           6: [10, [.8, .2, .2, .5], [.9, .1, .1, .5]],
+                           7: [10, [.7, .4, .5, .6], [.8, .2, .8, .7]],
+                           8: [9, [.5, .5, .4, .5], [.6, .4, .5, .3]],
+                           9: [10, [.6, .4, .4, .7], [.9, .1, .1, .9]],
+                           10: [10, [.1, .5, .2, .1], [.7, .2, .2, .5]],
+                           11: [10, [.2, .4, .5, .2], [.3, .5, .4, .2]],
+                           12: [10, [.6, .2, .2, .4], [.8, .1, .1, .4]],
+                           13: [10, [.6, .4, .7, .5], [.8, .2, .8, .5]],
+                           14: [9, [.1, .5, .5, .5], [.1, .4, .5, .4]],
+                           15: [10, [.1, .4, .5, .5], [.8, .3, .4, .9]],
+                           16: [10, [.6, .4, .7, .6], [.7, .2, .2, .5]],
+                           17: [10, [.5, .4, .6, .5], [.7, .2, .7, .6]],
+                           18: [10, [.7, .4, .2, .7], [.9, .1, .1, .9]],
+                           19: [10, [.7, .3, .3, .5], [.8, .1, .1, .6]],
+                           20: [10, [.6, .3, .2, .5], [.9, .1, .1, .6]]}
+    labels = dict()
+    for time_series_key in raw_training_labels:
+        time_series = raw_training_labels[time_series_key]
+        num_images = time_series[0]
+        increment = [(time_series[2][emotion_dimension_idx] - time_series[1][emotion_dimension_idx]) / num_images for emotion_dimension_idx in range(EMOTION_DIMENSION_COUNT)]
+        labels[time_series_key] = [[increment[label_idx]*image_idx + (time_series[1][label_idx]) for label_idx in range(EMOTION_DIMENSION_COUNT)] for image_idx in range(num_images)]
+
+    return labels
+
 
 def get_training_label_array():
     raw_training_labels = get_raw_training_labels()

svr_plus_tdnn/data.py

@@ -8,7 +8,7 @@
 
 def get_image_feature_vector_array():
     feature = Feature()
-    root_directory = "../images"
+    root_directory = "../data/cohn_kanade_images"
     features = list()
     for subfile in os.listdir(root_directory):
         if "DS_Store" in subfile:   continue
@@ -93,7 +93,7 @@ def get_time_delay_image_training_data(time_delay=2):
 
 def get_image_feature_vector_batches():
     feature = Feature()
-    root_directory = "../images"
+    root_directory = "../data/cohn_kanade_images"
     features = dict()
     idx = 1
     for subfile in os.listdir(root_directory):

svr_plus_tdnn/main.py

@@ -1,3 +1,5 @@
+import sys
+sys.path.append('../feature')
 from tdnn import TDNN
 from data import get_image_feature_vector_array, get_training_label_array, get_time_delay_training_data
 from regressionModel import RegressionModel