Skip to content

Commit c7dff82

Browse files
smistadsmistad
committed
try statement on import of torch tests
1 parent b2ac4f6 commit c7dff82

File tree

2 files changed

+131
-127
lines changed

2 files changed

+131
-127
lines changed

‎.github/workflows/CI-mac-x86_64.yml‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -187,5 +187,5 @@ jobs:
187187
run: |
188188
cd ${{github.workspace}}
189189
source venv/bin/activate
190-
pytest source/FAST/ -v -k 'not torch'
190+
pytest source/FAST/ -v
191191

‎source/FAST/Python/Tests/test_torch.py‎

Lines changed: 130 additions & 126 deletions
Original file line numberDiff line numberDiff line change
@@ -1,139 +1,143 @@
1-
import torch
2-
import fast
3-
import numpy as np
4-
import pytest
5-
6-
7-
def test_torch_tensor_fast_image_2d():
8-
types_to_test = [
9-
(torch.uint8, fast.TYPE_UINT8, 255),
10-
(torch.float32, fast.TYPE_FLOAT, 1),
11-
(torch.uint16, fast.TYPE_UINT16, 128),
12-
(torch.int16, fast.TYPE_INT16, 128)
13-
]
14-
width = 64
15-
height = 37
16-
17-
for torch_type, fast_type, scale in types_to_test:
18-
tensor = (torch.rand(height, width, dtype=torch.float32)*scale).to(torch_type)
19-
image = fast.Image.createFromTensor(tensor)
20-
21-
assert image.getWidth() == width
22-
assert image.getHeight() == height
23-
assert image.getNrOfChannels() == 1
24-
assert image.getDataType() == fast_type
25-
assert torch.equal(tensor, torch.tensor(np.asarray(image)).squeeze())
26-
27-
for channels in range(1, 5):
28-
tensor = (torch.rand(channels, height, width, dtype=torch.float32)*scale).to(torch_type)
1+
try:
2+
import torch
3+
import fast
4+
import numpy as np
5+
import pytest
6+
7+
8+
def test_torch_tensor_fast_image_2d():
9+
types_to_test = [
10+
(torch.uint8, fast.TYPE_UINT8, 255),
11+
(torch.float32, fast.TYPE_FLOAT, 1),
12+
(torch.uint16, fast.TYPE_UINT16, 128),
13+
(torch.int16, fast.TYPE_INT16, 128)
14+
]
15+
width = 64
16+
height = 37
17+
18+
for torch_type, fast_type, scale in types_to_test:
19+
tensor = (torch.rand(height, width, dtype=torch.float32)*scale).to(torch_type)
2920
image = fast.Image.createFromTensor(tensor)
3021

3122
assert image.getWidth() == width
3223
assert image.getHeight() == height
33-
assert image.getNrOfChannels() == channels
24+
assert image.getNrOfChannels() == 1
3425
assert image.getDataType() == fast_type
35-
assert torch.equal(tensor, torch.tensor(np.asarray(image)).permute((2, 0, 1)))
36-
37-
38-
def test_torch_tensor_fast_image_3d():
39-
types_to_test = [
40-
(torch.uint8, fast.TYPE_UINT8, 255),
41-
(torch.float32, fast.TYPE_FLOAT, 1),
42-
(torch.uint16, fast.TYPE_UINT16, 128),
43-
(torch.int16, fast.TYPE_INT16, 128)
44-
]
45-
width = 64
46-
height = 37
47-
depth = 42
48-
49-
for torch_type, fast_type, scale in types_to_test:
50-
tensor = (torch.rand(depth, height, width, dtype=torch.float32)*scale).to(torch_type)
51-
image = fast.Image.createFromTensor(tensor)
52-
53-
assert image.getWidth() == width
54-
assert image.getHeight() == height
55-
assert image.getDepth() == depth
56-
assert image.getNrOfChannels() == 1
57-
assert image.getDataType() == fast_type
58-
assert torch.equal(tensor, torch.tensor(np.asarray(image)).squeeze())
59-
60-
for channels in range(1, 5):
61-
tensor = (torch.rand(channels, depth, height, width, dtype=torch.float32)*scale).to(torch_type)
26+
assert torch.equal(tensor, torch.tensor(np.asarray(image)).squeeze())
27+
28+
for channels in range(1, 5):
29+
tensor = (torch.rand(channels, height, width, dtype=torch.float32)*scale).to(torch_type)
30+
image = fast.Image.createFromTensor(tensor)
31+
32+
assert image.getWidth() == width
33+
assert image.getHeight() == height
34+
assert image.getNrOfChannels() == channels
35+
assert image.getDataType() == fast_type
36+
assert torch.equal(tensor, torch.tensor(np.asarray(image)).permute((2, 0, 1)))
37+
38+
39+
def test_torch_tensor_fast_image_3d():
40+
types_to_test = [
41+
(torch.uint8, fast.TYPE_UINT8, 255),
42+
(torch.float32, fast.TYPE_FLOAT, 1),
43+
(torch.uint16, fast.TYPE_UINT16, 128),
44+
(torch.int16, fast.TYPE_INT16, 128)
45+
]
46+
width = 64
47+
height = 37
48+
depth = 42
49+
50+
for torch_type, fast_type, scale in types_to_test:
51+
tensor = (torch.rand(depth, height, width, dtype=torch.float32)*scale).to(torch_type)
6252
image = fast.Image.createFromTensor(tensor)
6353

6454
assert image.getWidth() == width
6555
assert image.getHeight() == height
6656
assert image.getDepth() == depth
67-
assert image.getNrOfChannels() == channels
57+
assert image.getNrOfChannels() == 1
6858
assert image.getDataType() == fast_type
69-
assert torch.equal(tensor, torch.tensor(np.asarray(image)).permute((3, 0, 1, 2)))
70-
71-
72-
def test_torch_tensor_to_image_exceptions():
73-
# Have to use createFromTensor not createFromArray
74-
tensor = torch.rand((1, 32, 32), dtype=torch.float32)
75-
with pytest.raises(ValueError):
76-
fast.Image.createFromArray(tensor)
77-
78-
# Only 1 dim
79-
tensor = torch.rand((7,), dtype=torch.float32)
80-
with pytest.raises(ValueError):
81-
fast.Image.createFromTensor(tensor)
82-
83-
# More than 4 dims
84-
tensor = torch.rand((7, 32, 13, 23, 54), dtype=torch.float32)
85-
with pytest.raises(ValueError):
86-
fast.Image.createFromTensor(tensor)
87-
88-
# Incorrect type
89-
tensor = torch.rand((1, 32, 32), dtype=torch.float64)
90-
with pytest.raises(TypeError):
91-
fast.Image.createFromTensor(tensor)
92-
with pytest.raises(ValueError):
93-
fast.Image.createFromTensor('')
94-
95-
96-
def test_torch_tensor_to_tensor_exceptions():
97-
# Incorrect type
98-
tensor = torch.rand((1, 32, 32), dtype=torch.float32)
99-
with pytest.raises(ValueError):
100-
fast.Tensor.createFromArray(tensor)
101-
with pytest.raises(ValueError):
102-
fast.Tensor.createFromTensor('')
103-
104-
105-
def test_torch_tensor_fast_tensor():
106-
types_to_test = [
107-
(torch.uint8, 255),
108-
(torch.int8, 127),
109-
(torch.float32, 1),
110-
(torch.float64, 1),
111-
(torch.uint16, 128),
112-
(torch.int16, 128)
113-
]
114-
for type, scale in types_to_test:
115-
shape = (23,)
116-
tensor = (torch.rand(shape)*scale).to(type)
117-
fast_tensor = fast.Tensor.createFromTensor(tensor)
59+
assert torch.equal(tensor, torch.tensor(np.asarray(image)).squeeze())
60+
61+
for channels in range(1, 5):
62+
tensor = (torch.rand(channels, depth, height, width, dtype=torch.float32)*scale).to(torch_type)
63+
image = fast.Image.createFromTensor(tensor)
64+
65+
assert image.getWidth() == width
66+
assert image.getHeight() == height
67+
assert image.getDepth() == depth
68+
assert image.getNrOfChannels() == channels
69+
assert image.getDataType() == fast_type
70+
assert torch.equal(tensor, torch.tensor(np.asarray(image)).permute((3, 0, 1, 2)))
71+
72+
73+
def test_torch_tensor_to_image_exceptions():
74+
# Have to use createFromTensor not createFromArray
75+
tensor = torch.rand((1, 32, 32), dtype=torch.float32)
76+
with pytest.raises(ValueError):
77+
fast.Image.createFromArray(tensor)
78+
79+
# Only 1 dim
80+
tensor = torch.rand((7,), dtype=torch.float32)
81+
with pytest.raises(ValueError):
82+
fast.Image.createFromTensor(tensor)
83+
84+
# More than 4 dims
85+
tensor = torch.rand((7, 32, 13, 23, 54), dtype=torch.float32)
86+
with pytest.raises(ValueError):
87+
fast.Image.createFromTensor(tensor)
88+
89+
# Incorrect type
90+
tensor = torch.rand((1, 32, 32), dtype=torch.float64)
91+
with pytest.raises(TypeError):
92+
fast.Image.createFromTensor(tensor)
93+
with pytest.raises(ValueError):
94+
fast.Image.createFromTensor('')
95+
96+
97+
def test_torch_tensor_to_tensor_exceptions():
98+
# Incorrect type
99+
tensor = torch.rand((1, 32, 32), dtype=torch.float32)
100+
with pytest.raises(ValueError):
101+
fast.Tensor.createFromArray(tensor)
102+
with pytest.raises(ValueError):
103+
fast.Tensor.createFromTensor('')
104+
105+
106+
def test_torch_tensor_fast_tensor():
107+
types_to_test = [
108+
(torch.uint8, 255),
109+
(torch.int8, 127),
110+
(torch.float32, 1),
111+
(torch.float64, 1),
112+
(torch.uint16, 128),
113+
(torch.int16, 128)
114+
]
115+
for type, scale in types_to_test:
116+
shape = (23,)
117+
tensor = (torch.rand(shape)*scale).to(type)
118+
fast_tensor = fast.Tensor.createFromTensor(tensor)
119+
assert fast_tensor.getShape().getDimensions() == len(shape)
120+
assert fast_tensor.getShape().getAll()[0] == shape[0]
121+
assert torch.equal(tensor.to(torch.float32), torch.tensor(np.array(fast_tensor)))
122+
123+
shape = (23,1,23,67)
124+
tensor = (torch.rand(shape)*scale).to(type)
125+
fast_tensor = fast.Tensor.createFromTensor(tensor)
126+
assert fast_tensor.getShape().getDimensions() == len(shape)
127+
for i in range(len(shape)):
128+
assert fast_tensor.getShape().getAll()[i] == shape[i]
129+
assert torch.equal(tensor.to(torch.float32), torch.tensor(np.array(fast_tensor)))
130+
131+
132+
def test_torch_tensor_fast_tensor_channels_last_conversion():
133+
shape = (2, 32, 40)
134+
tensor = torch.rand(shape, dtype=torch.float32)
135+
fast_tensor = fast.Tensor.createFromTensor(tensor, convertToChannelsLast=True)
118136
assert fast_tensor.getShape().getDimensions() == len(shape)
119-
assert fast_tensor.getShape().getAll()[0] == shape[0]
120-
assert torch.equal(tensor.to(torch.float32), torch.tensor(np.array(fast_tensor)))
137+
assert fast_tensor.getShape().getAll()[0] == shape[1]
138+
assert fast_tensor.getShape().getAll()[1] == shape[2]
139+
assert fast_tensor.getShape().getAll()[2] == shape[0]
140+
assert torch.equal(tensor.permute((1,2,0)), torch.tensor(np.array(fast_tensor)))
121141

122-
shape = (23,1,23,67)
123-
tensor = (torch.rand(shape)*scale).to(type)
124-
fast_tensor = fast.Tensor.createFromTensor(tensor)
125-
assert fast_tensor.getShape().getDimensions() == len(shape)
126-
for i in range(len(shape)):
127-
assert fast_tensor.getShape().getAll()[i] == shape[i]
128-
assert torch.equal(tensor.to(torch.float32), torch.tensor(np.array(fast_tensor)))
129-
130-
131-
def test_torch_tensor_fast_tensor_channels_last_conversion():
132-
shape = (2, 32, 40)
133-
tensor = torch.rand(shape, dtype=torch.float32)
134-
fast_tensor = fast.Tensor.createFromTensor(tensor, convertToChannelsLast=True)
135-
assert fast_tensor.getShape().getDimensions() == len(shape)
136-
assert fast_tensor.getShape().getAll()[0] == shape[1]
137-
assert fast_tensor.getShape().getAll()[1] == shape[2]
138-
assert fast_tensor.getShape().getAll()[2] == shape[0]
139-
assert torch.equal(tensor.permute((1,2,0)), torch.tensor(np.array(fast_tensor)))
142+
except ImportError:
143+
print('Torch not installed. Not able to run tests.')

0 commit comments

Comments
 (0)