def main():
try:
random.seed(42)
ins_size = 20
step = 4
label_option = 100
#training data
trainDS = trainds.read_data_sets(ins_size, step, label_option);
classifier = skflow.TensorFlowDNNClassifier(
hidden_units =[500,1000,500],
n_classes = label_option,
batch_size = 200000,
steps = 1,
learning_rate = 0.01)
classifier.fit(trainDS.train.images, np.argmax(trainDS.train.labels, axis = 1))
data_filename = 'skflow.txt'
start_time = time.time()
number = np.zeros((41,4))
index = -1
for det in range(1,8):
if (det!=6):
num_n = 6
else:
num_n = 5
for n in range(0,num_n):
index = index + 1
temp = np.zeros((3,1))
for angle in range(1,4):
filename = '../../images/detector_' + str(det) + '_no_' + str(n) \
+ '_angle_' + str(angle) + '.jpg'
elapse_time = (time.time()-start_time)
if(elapse_time >= 1):
remain_time = elapse_time/(index*3+angle-1)*41*3-elapse_time
print 'Processing .. ' + filename \
+ time.strftime(" %H:%M:%S", time.gmtime(elapse_time)) \
+ ' has past. ' + 'Remaining time: ' \
+ time.strftime(" %H:%M:%S", time.gmtime(remain_time))
else:
print 'Processing .. ' + filename \
+ time.strftime(" %H:%M:%S", time.gmtime(elapse_time)) \
+ ' has past'
img = io.imread(filename)
testDS = testds.read_data_sets(img, ins_size, step)
batch_num = testDS.test.ylength
batch_size = testDS.test.xlength
y = classifier.predict(testDS.test.images)
temp[angle-1] = np.sum(y)
number[index,1] = np.mean(y)
number[index,2] = np.std(y)
manual_count = np.array([1,27,40,79,122,160,1,18,28,42,121,223,0,11,24,46,\
142,173,3,19,23,76,191,197,0,15,24,45,91,152,0,\
16,27,34,88,0,9,12,69,104,123])
number[:,0] = manual_count.T
number.tofile(data_filename,sep=" ")
sess = tf.Session();
_y = tf.placeholder(tf.float32, [None], name="y")
p = tf.reshape(tf.to_float(_y),[batch_num, batch_size])
test_result = tf.reshape(p, [1, batch_num, batch_size, 1])
_test_label = tf.image_summary("Test Label", test_result, max_images=1)
merged_image = tf.merge_summary([_test_label])
writer = tf.train.SummaryWriter("/tmp/neuralnetwork_logs", \
sess.graph.as_graph_def(add_shapes=True))
init = tf.initialize_all_variables()
sess.run(init)
feed = {_y:y}
result_image = sess.run([merged_image, test_result], feed_dict=feed)
writer.add_summary(result_image[0])
sess.close()
except KeyboardInterrupt:
print "Shutdown requested... exiting"
except Exception:
traceback.print_exc(file=sys.stdout)
sys.exit(0)
def main():
try:
ins_size = 20
step = 4
label_option = 100
sess = tf.Session()
#training data
trainDS = trainds.read_data_sets(ins_size, step,label_option);
#testing data
testfile = '../../images/detector_2_no_5_angle_2.jpg'
img = io.imread(testfile)
testDS = testds.read_data_sets(img, ins_size, step)
batch_num = testDS.test.ylength
batch_size = testDS.test.xlength
# Create the model
x = tf.placeholder(tf.float32, shape=[None, ins_size**2*3], name = "x_input")
y_ = tf.placeholder(tf.float32, [None, label_option], name="y_input")
batch_index = tf.placeholder(tf.int32, shape = [], name = "batch_index")
W = tf.Variable(tf.zeros([(ins_size**2*3),label_option]), name = "weights")
b = tf.Variable(tf.zeros([label_option]), name = "bias")
p_train = tf.Variable(tf.ones([batch_num,batch_size]), name = "probability")
p_test = tf.Variable(tf.ones([batch_num,batch_size]), name = "probability")
with tf.name_scope("Wx_b") as scope:
y = tf.nn.softmax(tf.matmul(x, W) + b)
#Add summary ops to collect data
w_hist = tf.histogram_summary("weights", W)
b_hist = tf.histogram_summary("biases", b)
y_hist = tf.histogram_summary("y", y)
# Define loss and optimizer
with tf.name_scope("xent") as scope:
cross_entropy = -tf.reduce_sum(y_*tf.log(y))
ce_summ = tf.scalar_summary("cross entropy", cross_entropy)
with tf.name_scope("train") as scope:
p_train_batch = tf.reshape(tf.to_float(tf.argmax(y_,1)),[1, batch_size])
update_op_train = tf.scatter_update(p_train,
tf.expand_dims(batch_index%batch_num, 0),p_train_batch)
train_step = tf.train.GradientDescentOptimizer(0.00001).minimize(cross_entropy)
with tf.name_scope("test") as scope:
p_test_batch = tf.reshape(tf.to_float(tf.argmax(y,1)),[1, batch_size])
update_op_test = tf.scatter_update(p_test,
tf.expand_dims(batch_index, 0),p_test_batch)
#tf.expand_dims(p_test_batch, 0))
#Image Summary:
train_label = tf.reshape(p_train, [1, batch_num, batch_size, 1])
test_image = tf.reshape(tf.convert_to_tensor(img, dtype=tf.float32), \
[1, img.shape[0], img.shape[1], 3])
test_result = tf.reshape(p_test, [1, batch_num, batch_size, 1])
_train_label = tf.image_summary("Train Label", train_label, max_images=1)
_test_image = tf.image_summary("Test Image", test_image, max_images=1)
_test_label = tf.image_summary("Test Label", test_result, max_images=1)
merged_train = tf.merge_summary([ \
w_hist, b_hist, y_hist, ce_summ])
merged_image = tf.merge_summary([ \
_train_label, _test_image, _test_label])
writer = tf.train.SummaryWriter("/tmp/neuralnetwork_logs", \
sess.graph.as_graph_def(add_shapes=True))
init = tf.initialize_all_variables()
sess.run(init)
#train
for i in range(batch_num*1000):
batch_xs, batch_ys = trainDS.train.next_batch(batch_size)
feed = {x: batch_xs, y_: batch_ys, batch_index: i}
sess.run([update_op_train, train_step], feed_dict=feed)
if i % 100 == 0: # Record summary data
if i % 1000 == 0:
print i
result = sess.run([merged_train, \
update_op_train, train_step], feed_dict=feed)
writer.add_summary(result[0], i)
else:
sess.run([update_op_train, train_step], feed_dict=feed)
#sess.run(tf.Print(W,[W,b]))
#test
for i in range(batch_num):
batch_xs = testDS.test.next_batch(batch_size)
feed = {x: batch_xs, batch_index: i, y_: np.ones((batch_size, label_option))}
result = sess.run(update_op_test, feed_dict=feed)
result_image = sess.run(merged_image)
writer.add_summary(result_image, batch_num)
sess.close()
except KeyboardInterrupt:
print "Shutdown requested... exiting"
except Exception:
traceback.print_exc(file=sys.stdout)
sys.exit(0)
