CoreML中Python Tensorflow模型的图像输入

Question

I'm implementing a simple image classification model with tensor flow (python).

Here's my image preprocessing:

import glob
for filename in glob.glob('/Volumes/G-DRIVE mobile USB-C/traan/*.jpeg'): #assuming jpeg
    im=Image.open(filename)
    im = im.resize((150,120), Image.ANTIALIAS)
    print(im.size)
    training_images.append(im)

And here's my very simple model:

model = keras.Sequential([
    keras.layers.Flatten(input_shape=(120, 150, 3)),
    keras.layers.Dense(512, activation=tf.nn.relu),
    keras.layers.Dense(256, activation=tf.nn.relu),
    keras.layers.Dense(128, activation=tf.nn.relu),
    keras.layers.Dense(10, activation=tf.nn.softmax)
])

I want to load this model to CoreML something like this,

import coremltools

modelCoreML = coremltools.converters.tensorflow.convert(model, input_feature, output_feature)
modelCoreML.save("Model.mlmodel")

But how do I do this where I can input an image, and not a numpy stack? Should I process the image and turn it into the right format in the app itself, then put it in the model? How would I do this?

Answer 1

You need to supply the image_input_names argument to coremltools.converters.keras.convert() so that coremltools knows which inputs should be treated as images. This is explained in the documentation .