double 3x300x300 to Color 300x300 about coremlhelpers HOT 10 CLOSED

I have found a similiar issue from you https://stackoverflow.com/questions/44529869/converting-uiimage-to-mlmultiarray-for-keras-model/44538525#44538525
On the other hand, my output is now "Multiarray (double 4x1917x1)" because input data is in a wrong type, and yours is "Multiarray (double)",so do I need to convert the output data or will it be solved just fixing the input data ?

from coremlhelpers.

I have already tried input conversion but output still is wrong :(

from coremlhelpers.

This happens because Core ML does not understand that it has to interpret your input as an image. You need to tell coremltools that your input should be treated as an image.

from coremlhelpers.

Thanks,
And how can I do this? while conversion from TF to Core Ml? could you show me a sample, please?
Thanks

from coremlhelpers.

Is it the same that you explained here? https://stackoverflow.com/questions/48773397/keras-image-classifier-for-coreml-image-input-asks-multiarray-instead-of-image

from coremlhelpers.

I haven't used the TF converter, so no idea how to do it there.

from coremlhelpers.

I mean CoreML tools from keras-python, sorry

from coremlhelpers.

I read this from you "When converting your model to Core ML you also need to add input_names='camera frame' when you call coremltools.converters.keras.convert(), in addition to image_input_names='camera frame'. Otherwise coremltools won't see this input and doesn't know you want to treat it as an image."

from coremlhelpers.

Yes that is correct.

from coremlhelpers.

@periscopiojoe If you are still facing this issue, the tf-coreml converter has image_input_name variable to be set for telling the converter that input is an image. For more infor, check this example notebook: https://github.com/tf-coreml/tf-coreml/blob/master/examples/style_transfer_example.ipynb

from coremlhelpers.