如何在 Jetson Xavier 上的 Numba 中使用 OpenCV？

Question

我试图在一些函数中使用 OpenCV，我用一些 Numba 装饰器（例如nopython=True, parallel=True）注释了这些函数。我在 Jetson Xavier 上运行它，它是用Nvidia SDK Manager刷新的。

代码是这样的：

@jit(nopython=True, cache=True, parallel=True)
def decompress(data):
  result = list()
  for d in data:
    cv2_image = cv2.imdecode(d, cv2.IMREAD_COLOR)
    image = np.array(cv2_image)
    result.append(image)
 return result

但我收到一个错误：

TypingError: Failed in nopython mode pipeline (step: nopython frontend)
Unknown attribute 'resize' of type Module(<module 'cv2' from '/usr/lib/python2.7/dist-packages/cv2.so'>)

File "./my-script.py", line 297:
            def decompress(data):
                <source elided>
                    cv2_image = cv2.imdecode(d, cv2.IMREAD_COLOR)
                    ^

[1] During: typing of get attribute at /ssd/catkin_workspace/src/semantics-ros-wrapper/src/semseg.py (297)

File "./my-script.py", line 297:
            def decompress(data):
                <source elided>
                    cv2_image = cv2.imdecode(d, cv2.IMREAD_COLOR)
                    ^

This is not usually a problem with Numba itself but instead often caused by
the use of unsupported features or an issue in resolving types.

To see Python/NumPy features supported by the latest release of Numba visit:
http://numba.pydata.org/numba-doc/dev/reference/pysupported.html
and
http://numba.pydata.org/numba-doc/dev/reference/numpysupported.html

For more information about typing errors and how to debug them visit:
http://numba.pydata.org/numba-doc/latest/user/troubleshoot.html#my-code-doesn-t-compile

If you think your code should work with Numba, please report the error message
and traceback, along with a minimal reproducer at:
https://github.com/numba/numba/issues/new

甚至可以将 OpenCV 与 Numba 一起使用吗？

我正在使用：python2.7 和 numba-0.44.0。

Answer 1

Numba 尚不支持 OpenCV。如果您仍然希望它在您的函数中的 numpy 数组上运行，您可以设置nopython=False

这意味着您也将无法设置 parallel=True

这是来自Numba 用户手册：

Numba 有两种编译模式：nopython 模式和对象模式。前者产生更快的代码，但有一些限制，可以迫使 Numba 回退到后者。要防止 Numba 回退，而是引发错误，请传递 nopython=True。