1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
|
#!/usr/bin/env python
import numpy as np
import cv2 as cv
import os
import sys
import unittest
from tests_common import NewOpenCVTests
try:
if sys.version_info[:2] < (3, 0):
raise unittest.SkipTest('Python 2.x is not supported')
openvino_is_available = True
try:
from openvino.runtime import Core, Type, Layout, PartialShape
from openvino.preprocess import ResizeAlgorithm, PrePostProcessor
except ImportError:
openvino_is_available = False
def skip_if_openvino_not_available():
if not openvino_is_available:
raise unittest.SkipTest("OpenVINO isn't available from python.")
class AgeGenderOV:
def __init__(self, model_path, bin_path, device):
self.device = device
self.core = Core()
self.model = self.core.read_model(model_path, bin_path)
def reshape(self, new_shape):
self.model.reshape(new_shape)
def cfgPrePostProcessing(self, pp_callback):
ppp = PrePostProcessor(self.model)
pp_callback(ppp)
self.model = ppp.build()
def apply(self, in_data):
compiled_model = self.core.compile_model(self.model, self.device)
infer_request = compiled_model.create_infer_request()
results = infer_request.infer(in_data)
ov_age = results['age_conv3'].squeeze()
ov_gender = results['prob'].squeeze()
return ov_age, ov_gender
class AgeGenderGAPI:
tag = 'age-gender-net'
def __init__(self, model_path, bin_path, device):
g_in = cv.GMat()
inputs = cv.GInferInputs()
inputs.setInput('data', g_in)
# TODO: It'd be nice to pass dict instead.
# E.g cv.gapi.infer("net", {'data': g_in})
outputs = cv.gapi.infer(AgeGenderGAPI.tag, inputs)
age_g = outputs.at("age_conv3")
gender_g = outputs.at("prob")
self.comp = cv.GComputation(cv.GIn(g_in), cv.GOut(age_g, gender_g))
self.pp = cv.gapi.ov.params(AgeGenderGAPI.tag, \
model_path, bin_path, device)
def apply(self, in_data):
compile_args = cv.gapi.compile_args(cv.gapi.networks(self.pp))
gapi_age, gapi_gender = self.comp.apply(cv.gin(in_data), compile_args)
gapi_gender = gapi_gender.squeeze()
gapi_age = gapi_age.squeeze()
return gapi_age, gapi_gender
class test_gapi_infer_ov(NewOpenCVTests):
def test_age_gender_infer_image(self):
skip_if_openvino_not_available()
root_path = '/omz_intel_models/intel/age-gender-recognition-retail-0013/FP32/age-gender-recognition-retail-0013'
model_path = self.find_file(root_path + '.xml', [os.environ.get('OPENCV_DNN_TEST_DATA_PATH')], required=False)
bin_path = self.find_file(root_path + '.bin', [os.environ.get('OPENCV_DNN_TEST_DATA_PATH')], required=False)
device_id = 'CPU'
img_path = self.find_file('cv/face/david2.jpg', [os.environ.get('OPENCV_TEST_DATA_PATH')])
img = cv.imread(img_path)
# OpenVINO
def preproc(ppp):
ppp.input().model().set_layout(Layout("NCHW"))
ppp.input().tensor().set_element_type(Type.u8) \
.set_spatial_static_shape(img.shape[0], img.shape[1]) \
.set_layout(Layout("NHWC"))
ppp.input().preprocess().resize(ResizeAlgorithm.RESIZE_LINEAR)
ref = AgeGenderOV(model_path, bin_path, device_id)
ref.cfgPrePostProcessing(preproc)
ov_age, ov_gender = ref.apply(np.expand_dims(img, 0))
# OpenCV G-API (No preproc required)
comp = AgeGenderGAPI(model_path, bin_path, device_id)
gapi_age, gapi_gender = comp.apply(img)
# Check
self.assertEqual(0.0, cv.norm(ov_gender, gapi_gender, cv.NORM_INF))
self.assertEqual(0.0, cv.norm(ov_age, gapi_age, cv.NORM_INF))
def test_age_gender_infer_tensor(self):
skip_if_openvino_not_available()
root_path = '/omz_intel_models/intel/age-gender-recognition-retail-0013/FP32/age-gender-recognition-retail-0013'
model_path = self.find_file(root_path + '.xml', [os.environ.get('OPENCV_DNN_TEST_DATA_PATH')], required=False)
bin_path = self.find_file(root_path + '.bin', [os.environ.get('OPENCV_DNN_TEST_DATA_PATH')], required=False)
device_id = 'CPU'
img_path = self.find_file('cv/face/david2.jpg', [os.environ.get('OPENCV_TEST_DATA_PATH')])
img = cv.imread(img_path)
# Prepare data manually
tensor = cv.resize(img, (62, 62)).astype(np.float32)
tensor = np.transpose(tensor, (2, 0, 1))
tensor = np.expand_dims(tensor, 0)
# OpenVINO (No preproce required)
ref = AgeGenderOV(model_path, bin_path, device_id)
ov_age, ov_gender = ref.apply(tensor)
# OpenCV G-API (No preproc required)
comp = AgeGenderGAPI(model_path, bin_path, device_id)
gapi_age, gapi_gender = comp.apply(tensor)
# Check
self.assertEqual(0.0, cv.norm(ov_gender, gapi_gender, cv.NORM_INF))
self.assertEqual(0.0, cv.norm(ov_age, gapi_age, cv.NORM_INF))
def test_age_gender_infer_batch(self):
skip_if_openvino_not_available()
root_path = '/omz_intel_models/intel/age-gender-recognition-retail-0013/FP32/age-gender-recognition-retail-0013'
model_path = self.find_file(root_path + '.xml', [os.environ.get('OPENCV_DNN_TEST_DATA_PATH')], required=False)
bin_path = self.find_file(root_path + '.bin', [os.environ.get('OPENCV_DNN_TEST_DATA_PATH')], required=False)
device_id = 'CPU'
img_path1 = self.find_file('cv/face/david1.jpg', [os.environ.get('OPENCV_TEST_DATA_PATH')])
img_path2 = self.find_file('cv/face/david2.jpg', [os.environ.get('OPENCV_TEST_DATA_PATH')])
img1 = cv.imread(img_path1)
img2 = cv.imread(img_path2)
# img1 and img2 have the same size
batch_img = np.array([img1, img2])
# OpenVINO
def preproc(ppp):
ppp.input().model().set_layout(Layout("NCHW"))
ppp.input().tensor().set_element_type(Type.u8) \
.set_spatial_static_shape(img1.shape[0], img2.shape[1]) \
.set_layout(Layout("NHWC"))
ppp.input().preprocess().resize(ResizeAlgorithm.RESIZE_LINEAR)
ref = AgeGenderOV(model_path, bin_path, device_id)
ref.reshape(PartialShape([2, 3, 62, 62]))
ref.cfgPrePostProcessing(preproc)
ov_age, ov_gender = ref.apply(batch_img)
# OpenCV G-API
comp = AgeGenderGAPI(model_path, bin_path, device_id)
comp.pp.cfgReshape([2, 3, 62, 62]) \
.cfgInputModelLayout("NCHW") \
.cfgInputTensorLayout("NHWC") \
.cfgResize(cv.INTER_LINEAR)
gapi_age, gapi_gender = comp.apply(batch_img)
# Check
self.assertEqual(0.0, cv.norm(ov_gender, gapi_gender, cv.NORM_INF))
self.assertEqual(0.0, cv.norm(ov_age, gapi_age, cv.NORM_INF))
def test_age_gender_infer_planar(self):
skip_if_openvino_not_available()
root_path = '/omz_intel_models/intel/age-gender-recognition-retail-0013/FP32/age-gender-recognition-retail-0013'
model_path = self.find_file(root_path + '.xml', [os.environ.get('OPENCV_DNN_TEST_DATA_PATH')], required=False)
bin_path = self.find_file(root_path + '.bin', [os.environ.get('OPENCV_DNN_TEST_DATA_PATH')], required=False)
device_id = 'CPU'
img_path = self.find_file('cv/face/david2.jpg', [os.environ.get('OPENCV_TEST_DATA_PATH')])
img = cv.imread(img_path)
planar_img = np.transpose(img, (2, 0, 1))
planar_img = np.expand_dims(planar_img, 0)
# OpenVINO
def preproc(ppp):
ppp.input().tensor().set_element_type(Type.u8) \
.set_spatial_static_shape(img.shape[0], img.shape[1])
ppp.input().preprocess().resize(ResizeAlgorithm.RESIZE_LINEAR)
ref = AgeGenderOV(model_path, bin_path, device_id)
ref.cfgPrePostProcessing(preproc)
ov_age, ov_gender = ref.apply(planar_img)
# OpenCV G-API
comp = AgeGenderGAPI(model_path, bin_path, device_id)
comp.pp.cfgResize(cv.INTER_LINEAR)
gapi_age, gapi_gender = comp.apply(planar_img)
# Check
self.assertEqual(0.0, cv.norm(ov_gender, gapi_gender, cv.NORM_INF))
self.assertEqual(0.0, cv.norm(ov_age, gapi_age, cv.NORM_INF))
except unittest.SkipTest as e:
message = str(e)
class TestSkip(unittest.TestCase):
def setUp(self):
self.skipTest('Skip tests: ' + message)
def test_skip():
pass
pass
if __name__ == '__main__':
NewOpenCVTests.bootstrap()
|
