Compile CoreML Models

Author: Joshua Z. Zhang, Kazutaka Morita

This article is an introductory tutorial to deploy CoreML models with Relay.

For us to begin with, coremltools module is required to be installed.

A quick solution is to install via pip

pip install -U coremltools --user

or please refer to official site https://github.com/apple/coremltools

import tvm
import tvm.relay as relay
from tvm.contrib.download import download_testdata
import coremltools as cm
import numpy as np
from PIL import Image

Load pretrained CoreML model

We will download and load a pretrained mobilenet classification network provided by apple in this example

model_url = 'https://docs-assets.developer.apple.com/coreml/models/MobileNet.mlmodel'
model_file = 'mobilenet.mlmodel'
model_path = download_testdata(model_url, model_file, module='coreml')
# Now you have mobilenet.mlmodel on disk
mlmodel = cm.models.MLModel(model_path)

Out:

File /workspace/.tvm_test_data/coreml/mobilenet.mlmodel exists, skip.

Load a test image

A single cat dominates the examples!

img_url = 'https://github.com/dmlc/mxnet.js/blob/master/data/cat.png?raw=true'
img_path = download_testdata(img_url, 'cat.png', module='data')
img = Image.open(img_path).resize((224, 224))
x = np.transpose(img, (2, 0, 1))[np.newaxis, :]

Out:

File /workspace/.tvm_test_data/data/cat.png exists, skip.

Compile the model on Relay

We should be familiar with the process right now.

target = 'cuda'
shape_dict = {'image': x.shape}

# Parse CoreML model and convert into Relay computation graph
mod, params = relay.frontend.from_coreml(mlmodel, shape_dict)

with relay.build_config(opt_level=3):
    graph, lib, params = relay.build(mod,
                                     target,
                                     params=params)

Execute on TVM

The process is no different from other example

from tvm.contrib import graph_runtime
ctx = tvm.gpu(0)
dtype = 'float32'
m = graph_runtime.create(graph, lib, ctx)
# set inputs
m.set_input('image', tvm.nd.array(x.astype(dtype)))
m.set_input(**params)
# execute
m.run()
# get outputs
tvm_output = m.get_output(0)
top1 = np.argmax(tvm_output.asnumpy()[0])

Look up synset name

Look up prediction top 1 index in 1000 class synset.

synset_url = ''.join(['https://gist.githubusercontent.com/zhreshold/',
                      '4d0b62f3d01426887599d4f7ede23ee5/raw/',
                      '596b27d23537e5a1b5751d2b0481ef172f58b539/',
                      'imagenet1000_clsid_to_human.txt'])
synset_name = 'imagenet1000_clsid_to_human.txt'
synset_path = download_testdata(synset_url, synset_name, module='data')
with open(synset_path) as f:
    synset = eval(f.read())
print('Top-1 id', top1, 'class name', synset[top1])

Out:

File /workspace/.tvm_test_data/data/imagenet1000_clsid_to_human.txt exists, skip.
Top-1 id 470 class name candle, taper, wax light

Total running time of the script: ( 0 minutes 14.779 seconds)

Gallery generated by Sphinx-Gallery