naokishibuya/car-traffic-sign-classification

Built and trained a deep neural network to classify traffic signs using Tensorflow

car-traffic-sign-classification's Issues

ValueError: No objects to concatenate

Hi I have downloaded all the Data files and insert into the car-traffic-sign-classification-master as GSTRB folder and trying to follow the steps in Notebook. But getting this ValueError: No objects to concatenate. I am also attaching the error pic. Thank You!!
url = https://drive.google.com/file/d/0B9l6wLTrMbENbkxUaEs0emkxYVE/view

Always gives an output

Hello sir,
I'm working on the same project, for my senior.
I trained my model and everything is working fine if I am using pictures of traffic signs, but when I used a picture of a dog, it shows me that it is a: Road work!
I tried other pictures too, they are showing me the same thing, wrong detection.
So, what it is doing is always giving an output, even though there is nothing to detect in the picture!
Any idea how to solve this problem?
I thought about making a threshold for the probability, but it won't work all the time.
Is there any special function/method to do this, since I am new to CNN and TF, I do not know all capabilities of it.

Got this error when running the code right under The First Performance.

TypeErrorTraceback (most recent call last)
in ()
1 with Session() as session:
2 functions = [loader]
----> 3 pipeline = build_pipeline(functions, session, make_network1(), make_adam(1.0e-3))
4 train_evaluate(pipeline)

in make_network1(input_shape)
3 def make_network1(input_shape=INPUT_SHAPE):
4 return (NeuralNetwork()
----> 5 .input(input_shape)
6 .conv([5, 5, 6])
7 .max_pool()

/home/linux/dsmlp/70/770/cg181fcp/final_project/pipeline/network.py in input(self, shape)
57 """
58 with self.variable_scope('input'):
---> 59 self.x = tf.placeholder(tf.float32, (None, shape),name='x')
60 self.y = tf.placeholder(tf.int64, (None), name='y')
61 self.is_training = tf.placeholder(tf.bool, name='is_training')

/opt/conda/lib/python2.7/site-packages/tensorflow/python/ops/array_ops.pyc in placeholder(dtype, shape, name)
1603 evaluated directly.
1604 """
-> 1605 return gen_array_ops._placeholder(dtype=dtype, shape=shape, name=name)
1606
1607

/opt/conda/lib/python2.7/site-packages/tensorflow/python/ops/gen_array_ops.pyc in _placeholder(dtype, shape, name)
3085 if shape is None:
3086 shape = None
-> 3087 shape = _execute.make_shape(shape, "shape")
3088 _ctx = _context.context()
3089 if _ctx.in_graph_mode():

/opt/conda/lib/python2.7/site-packages/tensorflow/python/eager/execute.pyc in make_shape(v, arg_name)
144 shape = tensor_shape.as_shape(v)
145 except TypeError as e:
--> 146 raise TypeError("Error converting %s to a TensorShape: %s." % (arg_name, e))
147 except ValueError as e:
148 raise ValueError("Error converting %s to a TensorShape: %s." % (arg_name,

TypeError: Error converting shape to a TensorShape: int() argument must be a string or a number, not 'tuple'.

My tensorflow version is 1.4.0-dev20171012.