mmdeploy/tests/test_apis/test_onnx2openvino.py

import os
import os.path as osp
import tempfile

import numpy as np
import pytest
import torch
import torch.nn as nn

from mmdeploy.apis.openvino import is_available

openvino_skip = not is_available()


@pytest.mark.skip(reason='This a not test class but a utility class.')
class TestModel(nn.Module):

    def __init__(self):
        super().__init__()

    def forward(self, x):
        return x * 0.5


def generate_onnx_file(model, export_img, onnx_file):
    with torch.no_grad():
        dynamic_axes = {
            'input': {
                0: 'batch',
                2: 'width',
                3: 'height'
            },
            'output': {
                0: 'batch'
            }
        }
        torch.onnx.export(
            model,
            export_img,
            onnx_file,
            output_names=['output'],
            input_names=['input'],
            keep_initializers_as_inputs=True,
            do_constant_folding=True,
            verbose=False,
            opset_version=11,
            dynamic_axes=dynamic_axes)
        assert osp.exists(onnx_file)


def get_outputs(pytorch_model, openvino_model_path, input):
    output_pytorch = pytorch_model(input).numpy()

    from mmdeploy.apis.openvino import OpenVINOWrapper
    openvino_model = OpenVINOWrapper(openvino_model_path)
    openvino_output = openvino_model({'input': input})['output']

    return output_pytorch, openvino_output


@pytest.mark.skipif(openvino_skip, reason='OpenVINO not avaiable')
def test_onnx2openvino():
    from mmdeploy.apis.openvino import get_output_model_file, onnx2openvino
    pytorch_model = TestModel().eval()
    export_img = torch.rand([1, 3, 8, 8])
    onnx_file = tempfile.NamedTemporaryFile(suffix='.onnx').name
    generate_onnx_file(pytorch_model, export_img, onnx_file)

    input_info = {'input': export_img.shape}
    output_names = ['output']
    openvino_dir = tempfile.TemporaryDirectory().name
    onnx2openvino(input_info, output_names, onnx_file, openvino_dir)
    openvino_model_path = get_output_model_file(onnx_file, openvino_dir)
    assert osp.exists(openvino_model_path), \
        'The file (.xml) for OpenVINO IR has not been created.'

    test_img = torch.rand([1, 3, 16, 16])
    output_pytorch, openvino_output = get_outputs(pytorch_model,
                                                  openvino_model_path,
                                                  test_img)
    assert np.allclose(output_pytorch, openvino_output), \
        'OpenVINO and PyTorch outputs are not the same.'


@pytest.mark.skipif(openvino_skip, reason='OpenVINO not avaiable')
def test_can_not_run_onnx2openvino_without_mo():
    current_environ = dict(os.environ)
    os.environ.clear()

    is_error = False
    try:
        from mmdeploy.apis.openvino import onnx2openvino
        onnx2openvino({}, ['output'], 'tmp.onnx', '/tmp')
    except RuntimeError:
        is_error = True

    os.environ.update(current_environ)
    assert is_error, \
        'The onnx2openvino script was launched without checking for MO.'


@pytest.mark.skipif(openvino_skip, reason='OpenVINO not avaiable')
def test_get_input_shape_from_cfg():
    from mmdeploy.apis.openvino import get_input_shape_from_cfg

    # Test with default value
    model_cfg = {}
    input_shape = get_input_shape_from_cfg(model_cfg)
    assert input_shape == [1, 3, 800, 1344], \
        'The function returned a different default shape.'

    # Test with config that contains the required data.
    height, width = 800, 1200
    model_cfg = {'test_pipeline': [{}, {'img_scale': (width, height)}]}
    input_shape = get_input_shape_from_cfg(model_cfg)
    assert input_shape == [1, 3, height, width], \
        'The shape in the config does not match the output shape.'