python: inconsistant behaviour with more than 512 channels

[berak]

System information (version)

• OpenCV => 4.5.2
• Operating System / Platform => Windows 64 Bit
• Compiler => python3.8

Detailed description

opencv's Mat is restricted to <= 512 channels, but numpy arrays can have any number

Steps to reproduce

>>> a = np.ones((10,10,600),np.uint8)
>>> cv2.add(a,a) # no problem ???
>>> cv2.copyMakeBorder(a,3,3,3,3,cv2.BORDER_REFLECT101)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
cv2.error: OpenCV(4.5.2-pre) C:\p\opencv\modules\core\src\copy.cpp:1026: error: (-215:Assertion failed) top >= 0 && bottom >= 0 && left >= 0 && right >= 0 && _src.dims() <= 2 in function 'copyMakeBorder'

as an unrelated sidenote, cv::Mat will silently wrap around, given more than 512 channels:

cv::Mat m(10,10,CV_8UC(600)); // no exception !
cout << m.channels() << endl;

88

which isn't nice, either.

Issue submission checklist

• I report the issue, it's not a question an error he
• I checked the problem with documentation, FAQ, open issues,
  forum.opencv.org, Stack Overflow, etc and have not found solution
• I updated to latest OpenCV version and the issue is still there
• There is reproducer code and related data files: videos, images, onnx, etc

[alalek]

With Python use cv.utils.dumpInputArray() to see converted parameter internals.

CV_8UC(...) is a macro. It is caller responsibility to check passed values. There is no problem with cv::Mat itself (as wrong value is passed).

[berak]

ah, that's indeed useful (we can see, that the dims() are changing, like the exeption in the 1st post):

>>> a = np.ones((10,10,600),np.uint8)
>>> cv2.utils.dumpInputArray(a)
'InputArray: empty()=false kind=0x00010000 flags=0x01010000 total(-1)=60000 dims(-1)=3 size(-1)=[10 10 600] type(-1)=CV_8UC1'
>>> b = np.ones((10,10,60),np.uint8)
>>> cv2.utils.dumpInputArray(b)
'InputArray: empty()=false kind=0x00010000 flags=0x01010000 total(-1)=100 dims(-1)=2 size(-1)=10x10 type(-1)=CV_8UC60'