rogerdettloff opened a new issue #18911:
URL: https://github.com/apache/incubator-mxnet/issues/18911
## Description
I've noticed that image transforms like `npx.image.random_flip_left_right`
and `npx.image.random_flip_top_bottom` sometimes produce an incorrect
result--although, sometimes they work correctly. Please see my example below...
### Error Message
No error message is produced.
## To Reproduce
```
>>> from mxnet import np, npx
>>> npx.set_np()
>>> img=np.arange(11*9).reshape(11,9,1)
>>> img[:,:,0] # have a look at our test image for reference.
array([[ 0., 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.]])
>>> img_flip_lr = npx.image.random_flip_left_right(img)
>>> img_flip_lr[:,:,0] # correctly flipped image
array([[ 8., 7., 6., 5., 4., 3., 2., 1., 0.],
[17., 16., 15., 14., 13., 12., 11., 10., 9.],
[26., 25., 24., 23., 22., 21., 20., 19., 18.],
[35., 34., 33., 32., 31., 30., 29., 28., 27.],
[44., 43., 42., 41., 40., 39., 38., 37., 36.],
[53., 52., 51., 50., 49., 48., 47., 46., 45.],
[62., 61., 60., 59., 58., 57., 56., 55., 54.],
[71., 70., 69., 68., 67., 66., 65., 64., 63.],
[80., 79., 78., 77., 76., 75., 74., 73., 72.],
[89., 88., 87., 86., 85., 84., 83., 82., 81.],
[98., 97., 96., 95., 94., 93., 92., 91., 90.]])
>>> img_flip_lr = npx.image.random_flip_left_right(img) # try again...
>>> img_flip_lr[:,:,0] # sometimes I get a messed-up result...
array([[ 8.0000000e+00, 7.0000000e+00, 6.0000000e+00, 5.0000000e+00,
4.4816047e-38, 3.0000000e+00, 2.0000000e+00, 1.0000000e+00,
0.0000000e+00],
[ 1.7000000e+01, 1.6000000e+01, 1.5000000e+01, 1.4000000e+01,
5.9144397e-38, 1.2000000e+01, 1.1000000e+01, 1.0000000e+01,
9.0000000e+00],
[ 2.6000000e+01, 2.5000000e+01, 2.4000000e+01, 2.3000000e+01,
7.3440759e-34, 2.1000000e+01, 2.0000000e+01, 1.9000000e+01,
1.8000000e+01],
[ 3.5000000e+01, 3.4000000e+01, 3.3000000e+01, 3.2000000e+01,
4.8542180e-32, 3.0000000e+01, 2.9000000e+01, 2.8000000e+01,
2.7000000e+01],
[ 4.4000000e+01, 4.3000000e+01, 4.2000000e+01, 4.1000000e+01,
-5.2694353e-17, 3.9000000e+01, 3.8000000e+01, 3.7000000e+01,
3.6000000e+01],
[ 5.3000000e+01, 5.2000000e+01, 5.1000000e+01, 5.0000000e+01,
9.3095263e-41, 4.8000000e+01, 4.7000000e+01, 4.6000000e+01,
4.5000000e+01],
[ 6.2000000e+01, 6.1000000e+01, 6.0000000e+01, 5.9000000e+01,
4.6286172e-38, 5.7000000e+01, 5.6000000e+01, 5.5000000e+01,
5.4000000e+01],
[ 7.1000000e+01, 7.0000000e+01, 6.9000000e+01, 6.8000000e+01,
1.9254146e-37, 6.6000000e+01, 6.5000000e+01, 6.4000000e+01,
6.3000000e+01],
[ 8.0000000e+01, 7.9000000e+01, 7.8000000e+01, 7.7000000e+01,
1.8514722e-37, 7.5000000e+01, 7.4000000e+01, 7.3000000e+01,
7.2000000e+01],
[ 8.9000000e+01, 8.8000000e+01, 8.7000000e+01, 8.6000000e+01,
1.0652919e-38, 8.4000000e+01, 8.3000000e+01, 8.2000000e+01,
8.1000000e+01],
[ 9.8000000e+01, 9.7000000e+01, 9.6000000e+01, 9.5000000e+01,
2.3658196e-37, 9.3000000e+01, 9.2000000e+01, 9.1000000e+01,
9.0000000e+01]])
```
The example above does not always produce messed-up results for me, but I
found that it seems more frequently messed-up when I Compose several sequential
transformations, like this:
```
>>> from mxnet.gluon.data.vision import transforms
>>> transformer = transforms.Compose([
... transforms.RandomFlipLeftRight(),
... transforms.RandomFlipTopBottom()
... ])
>>> transformer(img)[:,:,0]
array([[8.0000000e+00, 7.0000000e+00, 6.0000000e+00, 5.0000000e+00,
6.3878950e-04, 3.0000000e+00, 2.0000000e+00, 1.0000000e+00,
0.0000000e+00],
[1.7000000e+01, 1.6000000e+01, 1.5000000e+01, 1.4000000e+01,
4.7425812e+30, 1.2000000e+01, 1.1000000e+01, 1.0000000e+01,
9.0000000e+00],
[2.6000000e+01, 2.5000000e+01, 2.4000000e+01, 2.3000000e+01,
1.7657325e+22, 2.1000000e+01, 2.0000000e+01, 1.9000000e+01,
1.8000000e+01],
[3.5000000e+01, 3.4000000e+01, 3.3000000e+01, 3.2000000e+01,
7.5839171e+31, 3.0000000e+01, 2.9000000e+01, 2.8000000e+01,
2.7000000e+01],
[4.4000000e+01, 4.3000000e+01, 4.2000000e+01, 4.1000000e+01,
3.3809470e+03, 3.9000000e+01, 3.8000000e+01, 3.7000000e+01,
3.6000000e+01],
[5.3000000e+01, 5.2000000e+01, 5.1000000e+01, 5.0000000e+01,
2.2267046e-15, 4.8000000e+01, 4.7000000e+01, 4.6000000e+01,
4.5000000e+01],
[6.2000000e+01, 6.1000000e+01, 6.0000000e+01, 5.9000000e+01,
7.5033516e+28, 5.7000000e+01, 5.6000000e+01, 5.5000000e+01,
5.4000000e+01],
[7.1000000e+01, 7.0000000e+01, 6.9000000e+01, 6.8000000e+01,
4.7427375e+30, 6.6000000e+01, 6.5000000e+01, 6.4000000e+01,
6.3000000e+01],
[8.0000000e+01, 7.9000000e+01, 7.8000000e+01, 7.7000000e+01,
1.4589531e-19, 7.5000000e+01, 7.4000000e+01, 7.3000000e+01,
7.2000000e+01],
[8.9000000e+01, 8.8000000e+01, 8.7000000e+01, 8.6000000e+01,
4.7427375e+30, 8.4000000e+01, 8.3000000e+01, 8.2000000e+01,
8.1000000e+01],
[9.8000000e+01, 9.7000000e+01, 9.6000000e+01, 9.5000000e+01,
1.9209545e+31, 9.3000000e+01, 9.2000000e+01, 9.1000000e+01,
9.0000000e+01]])
>>> transformer(img)[:,:,0] # try again, and sometimes it works okay...
array([[90., 91., 92., 93., 94., 95., 96., 97., 98.],
[81., 82., 83., 84., 85., 86., 87., 88., 89.],
[72., 73., 74., 75., 76., 77., 78., 79., 80.],
[63., 64., 65., 66., 67., 68., 69., 70., 71.],
[54., 55., 56., 57., 58., 59., 60., 61., 62.],
[45., 46., 47., 48., 49., 50., 51., 52., 53.],
[36., 37., 38., 39., 40., 41., 42., 43., 44.],
[27., 28., 29., 30., 31., 32., 33., 34., 35.],
[18., 19., 20., 21., 22., 23., 24., 25., 26.],
[ 9., 10., 11., 12., 13., 14., 15., 16., 17.],
[ 0., 1., 2., 3., 4., 5., 6., 7., 8.]])
```
## Environment
```
----------Python Info----------
Version : 3.7.6
Compiler : MSC v.1916 64 bit (AMD64)
Build : ('tags/v3.7.6:43364a7ae0', 'Dec 19 2019 00:42:30')
Arch : ('64bit', 'WindowsPE')
------------Pip Info-----------
Version : 20.2.1
Directory :
D:\Projects\Combinati\dPCR-analysis\env-mxnet\lib\site-packages\pip
----------MXNet Info-----------
Version : 1.6.0
Directory :
D:\Projects\Combinati\dPCR-analysis\env-mxnet\lib\site-packages\mxnet
Num GPUs : 0
Hashtag not found. Not installed from pre-built package.
----------System Info----------
Platform : Windows-10-10.0.18362-SP0
system : Windows
node : roger-5577
release : 10
version : 10.0.18362
----------Hardware Info----------
machine : AMD64
processor : Intel64 Family 6 Model 158 Stepping 9, GenuineIntel
Name
Intel(R) Core(TM) i7-7700HQ CPU @ 2.80GHz
----------Network Test----------
Setting timeout: 10
Timing for MXNet: https://github.com/apache/incubator-mxnet, DNS: 0.0080
sec, LOAD: 1.1030 sec.
Timing for GluonNLP GitHub: https://github.com/dmlc/gluon-nlp, DNS: 0.0010
sec, LOAD: 0.5180 sec.
Timing for GluonNLP: http://gluon-nlp.mxnet.io, DNS: 0.0861 sec, LOAD:
0.3459 sec.
Timing for D2L: http://d2l.ai, DNS: 0.0520 sec, LOAD: 0.2090 sec.
Timing for D2L (zh-cn): http://zh.d2l.ai, DNS: 0.0190 sec, LOAD: 0.2664 sec.
Timing for FashionMNIST:
https://repo.mxnet.io/gluon/dataset/fashion-mnist/train-labels-idx1-ubyte.gz,
DNS: 0.0870 sec, LOAD: 0.4820 sec.
Timing for PYPI: https://pypi.python.org/pypi/pip, DNS: 0.0100 sec, LOAD:
0.5760 sec.
Error open Conda: https://repo.continuum.io/pkgs/free/, HTTP Error 403:
Forbidden, DNS finished in 0.006048440933227539 sec.
```
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
