working solutions

.gitignore

@@ -2,3 +2,5 @@ data
 .ipynb_checkpoints
 *.h5
 models
+*.swp
+*.p

keras_convnet.py

@@ -9,9 +9,9 @@
 from keras.layers import Convolution2D, MaxPooling2D, GlobalAveragePooling2D
 from keras.layers import Activation, Dropout, Dense, ELU, BatchNormalization
 
-with open('data/train.p', mode='rb') as f:
+with open('train.p', mode='rb') as f:
     train = pickle.load(f)
-with open('data/test.p', mode='rb') as f:
+with open('test.p', mode='rb') as f:
     test = pickle.load(f)
 
 X_train, X_val, y_train, y_val = train_test_split(train['features'], train['labels'], test_size=0.33, random_state=0)

tf_convnet.py

@@ -1,9 +1,75 @@
 """
 Start pure TensorFlow version.
-
-Does using TFSlim count?
 """
 import tensorflow as tf
 import pickle
 from sklearn.model_selection import train_test_split
 
+with open('train.p', mode='rb') as f:
+    train = pickle.load(f)
+with open('test.p', mode='rb') as f:
+    test = pickle.load(f)
+
+X_train, X_val, y_train, y_val = train_test_split(train['features'], train['labels'], test_size=0.33, random_state=0)
+X_test, y_test = test['features'], test['labels']
+
+X_train = X_train.astype('float32')
+X_val = X_val.astype('float32')
+X_test = X_test.astype('float32')
+
+# 0-255 -> 0-1
+X_train /= 255
+X_val /= 255
+X_test /= 255
+
+batch_size = 64
+# number of traffic signs
+n_classes = 43
+epochs = 10
+input_shape = X_train.shape[1:]
+
+feature_size = 32*32*3
+
+# tf.nn.conv2d(input, filter, strides, padding)
+W1 = tf.Variable(tf.random_normal((feature_size, 100), mean=0, stddev=0.01) , name="W1")
+W2 = tf.Variable(tf.random_normal((100, n_classes), mean=0, stddev=0.01) , name="W2")
+
+b1 = tf.Variable(tf.zeros((100)) , name="b1")
+b2 = tf.Variable(tf.zeros((n_classes)) , name="b2")
+
+features = tf.placeholder(tf.float32, (None, feature_size), name="features")
+labels = tf.placeholder(tf.int64, (None), name="labels")
+
+out1 = tf.matmul(features, W1) + b1
+out1 = tf.nn.relu(out1)
+logits = tf.matmul(out1, W2) + b2
+
+correct_prediction = tf.equal(tf.argmax(logits, 1), labels)
+accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
+
+loss = tf.reduce_mean(tf.nn.sparse_softmax_cross_entropy_with_logits(logits, labels), 0)
+optimizer = tf.train.AdamOptimizer()
+train_op = optimizer.minimize(loss)
+
+with tf.Session() as sess:
+	sess.run(tf.initialize_all_variables())
+	for i in range(epochs):
+		for offset in range(0, X_train.shape[0], batch_size):
+			end = offset + batch_size
+			batch_X = X_train[offset:end]
+			batch_y = y_train[offset:end]	
+			bs = batch_X.shape[0]
+			batch_X = batch_X.reshape(bs, feature_size)
+
+			l, _ = sess.run([loss, train_op], feed_dict={features: batch_X, labels: batch_y})
+
+		val_l, val_acc = sess.run([loss, accuracy], feed_dict={features: X_val.reshape(X_val.shape[0], feature_size), labels: y_val})
+		print("Validation	Loss =", val_l)
+		print("Validation	Accuracy =", val_acc)
+
+
+	test_l, test_acc = sess.run([loss, accuracy], feed_dict={features: X_test.reshape(X_test.shape[0], feature_size), labels: y_test})
+	print("Testing	Loss =", test_l)
+	print("Testing	Accuracy =", test_acc)
+
+