跟随教程，但 label-studio 安装会修改 numpy版本，引发广泛依赖的错误

[chenxinfeng4]

跟着 label-studio SAM 教程。在安装 label-studio 时，它会强制指定 numpy==1.21.6，继而引发pytorch/ cv2 等依赖的冲突。非常难受。

我踩坑了。于是必须重装 pytorch cv2 mmcv 等一系列的依赖库。所幸，最后搞定了依赖，也能跑通教程。

强烈建议在教程中想用户警告 ·label studio· 会修改numpy版本引发冲突。并且在教程中，先安装 label studio 再安装 pytorch 和 mmcv 等。

感谢

[mm-assistant[bot]]

We recommend using English or English & Chinese for issues so that we could have broader discussion.

[JimmyMa99]

感谢指出，后续将在重新上传的第一个视频中加入提醒。

[deandean82]

Maybe the problem about opencv "TypeError: 'numpy._DTypeMeta' object is not subscriptable" Fix cv2/typing/__init__.py line 68 - 83, it works for me

if numpy.lib.NumpyVersion(numpy.__version__) > "1.20.0" and sys.version_info >= (3, 9):
    NumPyArrayGeneric = 'numpy.ndarray[typing.Any, numpy.dtype[numpy.generic]]'
else:
    NumPyArrayGeneric = 'numpy.ndarray'


if numpy.lib.NumpyVersion(numpy.__version__) > "1.20.0" and sys.version_info >= (3, 9):
    NumPyArrayFloat32 = 'numpy.ndarray[typing.Any, numpy.dtype[numpy.float32]]'
else:
    NumPyArrayFloat32 = 'numpy.ndarray'


if numpy.lib.NumpyVersion(numpy.__version__) > "1.20.0" and sys.version_info >= (3, 9):
    NumPyArrayFloat64 = 'numpy.ndarray[typing.Any, numpy.dtype[numpy.float64]]'
else:
    NumPyArrayFloat64 = 'numpy.ndarray'