Add SpaceToDepth in frontend

[ruimashita]

Add SpaceToDepth in frontend

[CLAassistant]

All committers have signed the CLA.

[tjingrant]

@ruimashita looks like there's an additional CI error caused by SpaceToDepth:

ERROR: test_space_to_depth (frontend.test_node.TestNode)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/travis/build/onnx/onnx-tensorflow/test/frontend/test_node.py", line 86, in do_test_expected
    tf_output = sess.run(test_op, tf_feed_dict)
  File "/home/travis/virtualenv/python2.7.14/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 900, in run
    run_metadata_ptr)
  File "/home/travis/virtualenv/python2.7.14/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 1135, in _run
    feed_dict_tensor, options, run_metadata)
  File "/home/travis/virtualenv/python2.7.14/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 1316, in _do_run
    run_metadata)
  File "/home/travis/virtualenv/python2.7.14/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 1335, in _do_call
    raise type(e)(node_def, op, message)
InvalidArgumentError: Only NHWC data_format supported on CPU. Got NCHW
	 [[Node: SpaceToDepth = SpaceToDepth[T=DT_FLOAT, block_size=2, data_format="NCHW", _device="/job:localhost/replica:0/task:0/device:CPU:0"](_arg_in_0_0_0)]]
Caused by op u'SpaceToDepth', defined at:
  File "/opt/python/2.7.14/lib/python2.7/runpy.py", line 174, in _run_module_as_main
    "__main__", fname, loader, pkg_name)
  File "/opt/python/2.7.14/lib/python2.7/runpy.py", line 72, in _run_code
    exec code in run_globals
  File "/opt/python/2.7.14/lib/python2.7/unittest/__main__.py", line 12, in <module>
    main(module=None)
  File "/opt/python/2.7.14/lib/python2.7/unittest/main.py", line 95, in __init__
    self.runTests()
  File "/opt/python/2.7.14/lib/python2.7/unittest/main.py", line 232, in runTests
    self.result = testRunner.run(self.test)
  File "/opt/python/2.7.14/lib/python2.7/unittest/runner.py", line 151, in run
    test(result)
  File "/opt/python/2.7.14/lib/python2.7/unittest/suite.py", line 70, in __call__
    return self.run(*args, **kwds)
  File "/opt/python/2.7.14/lib/python2.7/unittest/suite.py", line 108, in run
    test(result)
  File "/opt/python/2.7.14/lib/python2.7/unittest/suite.py", line 70, in __call__
    return self.run(*args, **kwds)
  File "/opt/python/2.7.14/lib/python2.7/unittest/suite.py", line 108, in run
    test(result)
  File "/opt/python/2.7.14/lib/python2.7/unittest/suite.py", line 70, in __call__
    return self.run(*args, **kwds)
  File "/opt/python/2.7.14/lib/python2.7/unittest/suite.py", line 108, in run
    test(result)
  File "/opt/python/2.7.14/lib/python2.7/unittest/case.py", line 393, in __call__
    return self.run(*args, **kwds)
  File "/opt/python/2.7.14/lib/python2.7/unittest/case.py", line 329, in run
    testMethod()
  File "/home/travis/build/onnx/onnx-tensorflow/test/frontend/test_node.py", line 67, in do_test_expected
    test_op = tf_op(*tf_param_list, **attrs)
  File "/home/travis/virtualenv/python2.7.14/lib/python2.7/site-packages/tensorflow/python/ops/array_ops.py", line 2332, in space_to_depth
    return gen_array_ops.space_to_depth(input, block_size, data_format, name=name)
  File "/home/travis/virtualenv/python2.7.14/lib/python2.7/site-packages/tensorflow/python/ops/gen_array_ops.py", line 7638, in space_to_depth
    data_format=data_format, name=name)
  File "/home/travis/virtualenv/python2.7.14/lib/python2.7/site-packages/tensorflow/python/framework/op_def_library.py", line 787, in _apply_op_helper
    op_def=op_def)
  File "/home/travis/virtualenv/python2.7.14/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line 3392, in create_op
    op_def=op_def)
  File "/home/travis/virtualenv/python2.7.14/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line 1718, in __init__
    self._traceback = self._graph._extract_stack()  # pylint: disable=protected-access
InvalidArgumentError (see above for traceback): Only NHWC data_format supported on CPU. Got NCHW
	 [[Node: SpaceToDepth = SpaceToDepth[T=DT_FLOAT, block_size=2, data_format="NCHW", _device="/job:localhost/replica:0/task:0/device:CPU:0"](_arg_in_0_0_0)]]

[ruimashita]

@tjingrant
Basically, The problem is design of frontend test, I think.
The above error cause from tensorflow CPU implementation, the error don't cause on GPU. If I want to pass the NCHW test, I have to hard work to implement tensorflow itself.
To avoid these situation, Ideally frontend test should be aside from tensorflow implementation. Frontend test divide from backend, and then we should create new integrate test front and back, I think.

But, It is hard for me to re-design frontend test.
To resolove the error realistic, I rewrote SpaceToDepth and the test.
Please review it.

[fumihwh]

space_to_depth_name should be based on transpose_name
Or we should clearify the input is transposed. _T_STD_

[fumihwh]

Maybe could use get_perm_from_formats in backend.py

[fumihwh]

Also here.

[ruimashita]

@fumihwh
Thanks for comments.
I fixed.

[tjingrant]

@ruimashita thanks, we haven't really started to deal with the data format issue in the frontend, sorry that you have to worry about that issue..

[ruimashita]

@tjingrant
Thanks a lot for your leads and managements of this open source.