ValueError: tensorflow op ResizeBicubic is not supported

[thefiddler]

When converting the attached model I am getting an error that bicubic resampling is not supported.

python -m tf2onnx.convert --input model.pb --inputs image:0 --outputs segmentation:0 --fold_const

using tensorflow=1.12.0, onnx=1.4.1, opset=7, tfonnx=0.4.0/209e06
2019-01-29 11:40:48.744284: I tensorflow/tools/graph_transforms/transform_graph.cc:317] Applying fold_constants
2019-01-29 11:40:48.757588: I tensorflow/tools/graph_transforms/transform_graph.cc:317] Applying remove_attribute
2019-01-29 11:40:48.757800: I tensorflow/tools/graph_transforms/transform_graph.cc:317] Applying fold_batch_norms
2019-01-29 11:40:48.758288: I tensorflow/tools/graph_transforms/transform_graph.cc:317] Applying fold_old_batch_norms
2019-01-29 11:40:48.764933: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
Traceback (most recent call last):
  File "/usr/local/Cellar/python/3.7.2_1/Frameworks/Python.framework/Versions/3.7/lib/python3.7/runpy.py", line 193, in _run_module_as_main
    "__main__", mod_spec)
  File "/usr/local/Cellar/python/3.7.2_1/Frameworks/Python.framework/Versions/3.7/lib/python3.7/runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "~/.virtualenvs/onnx/lib/python3.7/site-packages/tf2onnx/convert.py", line 119, in <module>
    main()
  File "~/.virtualenvs/onnx/lib/python3.7/site-packages/tf2onnx/convert.py", line 102, in main
    inputs_as_nchw=args.inputs_as_nchw)
  File "~/.virtualenvs/onnx/lib/python3.7/site-packages/tf2onnx/tfonnx.py", line 2506, in process_tf_graph
    mapped_op, unmapped_op = tensorflow_onnx_mapping(g, continue_on_error, custom_op_handlers)
  File "~/.virtualenvs/onnx/lib/python3.7/site-packages/tf2onnx/tfonnx.py", line 2319, in tensorflow_onnx_mapping
    raise ValueError("tensorflow op " + op + " is not supported")
ValueError: tensorflow op ResizeBicubic is not supported

The model tries to resize the input image to 256x256 using bicubic interpolation.

[nbcsm]

as far as I know, bicubic interpolation is not supported by onnx so far.
@linkerzhang for awareness.

[codarag]

Hello,

I have the same issue when converting a TF model with opset 12 and tf2onnx 1.7.2, however as far as I know ONNX should support bicubic resampling since opset 10 (https://github.com/onnx/onnx/blob/master/docs/Operators.md#Resize).

Model conversion terminates fine, but I have this two error lines:

2021-01-22 09:48:20,789 - ERROR - Tensorflow op [StatefulPartitionedCall/model/resizing/resize/ResizeBicubic: ResizeBicubic] is not supported
2021-01-22 09:48:20,857 - ERROR - Unsupported ops: Counter({'ResizeBicubic': 1})

Is TF ResizeBicubic supported? Will be supported in future tf2onnx release?
Thanks

[guschmue]

Fair, we can add this.