Default MaxPoolingOp only supports NHWC on device type CPU

[lakefox]

Hi, I keep getting the same error "Default MaxPoolingOp only supports NHWC on device type CPU" every time I run it. I'm running it on a MacBook Pro so it's the only CPU, does thid work on CPU machines or only CPU/GPU?

Also sorry if this is a newbie question I am still learning TensorFlow.

Here's the full error message.

% python3.7 obj_detect_tracking.py --model_path obj_v3_model --version 3 --video_dir v1-val_testvideos \
--video_lst_file v1-val_testvideos.lst --frame_gap 1 --get_tracking
--tracking_dir test_track_out

0%| | 0/1 [00:02<?, ?it/s]
Traceback (most recent call last):
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1365, in _do_call
return fn(*args)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1350, in _run_fn
target_list, run_metadata)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1443, in _call_tf_sessionrun
run_metadata)
tensorflow.python.framework.errors_impl.InvalidArgumentError: Default MaxPoolingOp only supports NHWC on device type CPU
[[{{node pool0/max_pooling2d/MaxPool}}]]

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "obj_detect_tracking.py", line 647, in
sess_input, feed_dict=feed_dict)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 956, in run
run_metadata_ptr)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1180, in _run
feed_dict_tensor, options, run_metadata)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1359, in _do_run
run_metadata)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/client/session.py", line 1384, in _do_call
raise type(e)(node_def, op, message)
tensorflow.python.framework.errors_impl.InvalidArgumentError: Default MaxPoolingOp only supports NHWC on device type CPU
[[node pool0/max_pooling2d/MaxPool (defined at /Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/framework/ops.py:1748) ]]

Original stack trace for 'pool0/max_pooling2d/MaxPool':
File "obj_detect_tracking.py", line 517, in
model = get_model(args, args.gpuid_start, controller=args.controller)
File "/Users/masonwright/Desktop/objdection/od/models.py", line 110, in get_model
model = Mask_RCNN_FPN(config, gpuid=gpuid)
File "/Users/masonwright/Desktop/objdection/od/models.py", line 292, in init
self.build_forward()
File "/Users/masonwright/Desktop/objdection/od/models.py", line 479, in build_forward
use_se=config.use_se, use_resnext=config.use_resnext)
File "/Users/masonwright/Desktop/objdection/od/nn.py", line 895, in resnet_fpn_backbone
scope="pool0", data_format="NCHW")
File "/Users/masonwright/Desktop/objdection/od/nn.py", line 790, in MaxPooling
"channels_last" if data_format == "NHWC" else "channels_first")
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/util/deprecation.py", line 324, in new_func
return func(*args, **kwargs)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/layers/pooling.py", line 311, in max_pooling2d
return layer.apply(inputs)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/util/deprecation.py", line 324, in new_func
return func(*args, **kwargs)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/keras/engine/base_layer.py", line 1700, in apply
return self.call(inputs, *args, **kwargs)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/layers/base.py", line 548, in call
outputs = super(Layer, self).call(inputs, *args, **kwargs)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/keras/engine/base_layer.py", line 854, in call
outputs = call_fn(cast_inputs, *args, **kwargs)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/autograph/impl/api.py", line 234, in wrapper
return converted_call(f, options, args, kwargs)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/autograph/impl/api.py", line 439, in converted_call
return _call_unconverted(f, args, kwargs, options)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/autograph/impl/api.py", line 330, in _call_unconverted
return f(*args, **kwargs)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/keras/layers/pooling.py", line 248, in call
data_format=conv_utils.convert_data_format(self.data_format, 4))
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/ops/nn_ops.py", line 3815, in max_pool
name=name)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/ops/gen_nn_ops.py", line 5674, in max_pool
data_format=data_format, name=name)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/framework/op_def_library.py", line 794, in _apply_op_helper
op_def=op_def)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/util/deprecation.py", line 507, in new_func
return func(*args, **kwargs)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/framework/ops.py", line 3357, in create_op
attrs, op_def, compute_device)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/framework/ops.py", line 3426, in _create_op_internal
op_def=op_def)
File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/framework/ops.py", line 1748, in init
self._traceback = tf_stack.extract_stack()

[JunweiLiang]

Hi,

Yes currently the code needs tensorflow-gpu to run. To run it with CPU, you'll need to change a lot of code to make tensor dimensions from [N, C, H, W] to [N, H, W, C].
You can probably try it on colab with free GPUs.