mmcv1.x版本

FAQ文档

嘻嘻

open-mmlab

哈哈

卸载环境中的mmcv

重新下载

ImportError: cannot import name ‘Config’ from ‘mmcv’ (unknown location)

problem description

解决

在mmcv 2.0.0中，Config已移动到mmengine，进行如下修改即可：

from mmengine.config import Config

MMCV官网
参考链接
参考

ImportError: cannot import name ‘mkdir_or_exist’ from ‘mmcv.utils’ (unknown location)

解决

很简单，改为

from mmcv.utils.path import mkdir_or_exist

即可，mkdir_or_exist在utils.path中定义的

AttributeError: module ‘mmcv’ has no attribute ‘ProgressBar’

简单来说就是mmcv中没有这个函数，我的版本为mmcv-full 1.7

好吧，查看了下文档实际上有

from mmcv.utils import progressbar

然后直接用progressbar函数即可

很棒，报错

受不了了，一个没啥用的还一直报错。。。

解决

直接用tqdm替代了

# 获取目录下文件的数量
    total_files = len(os.listdir(origin_label_path))

    # 使用tqdm创建进度条对象
    with tqdm(total=total_files) as pbar:
        for label_name in os.listdir(origin_label_path):
            image_objects = wwtool.simpletxt_parse(os.path.join(origin_label_path, label_name))
            filtered_objects = []
            for image_object in image_objects:
                if convert_classes[image_object['label']] == None:
                    filter_count += 1
                    continue
                else:
                    image_object['label'] = convert_classes[image_object['label']]
                    filtered_objects.append(image_object)

            if len(filtered_objects) > 0:
                img = cv2.imread(os.path.join(origin_image_path, os.path.splitext(label_name)[0] + image_format2))
                save_image_file = os.path.join(filtered_image_path, os.path.splitext(label_name)[0] + '.png')
                # print("Save image file: ", save_image_file)
                cv2.imwrite(save_image_file, img)
                wwtool.simpletxt_dump(filtered_objects, os.path.join(filtered_label_path, os.path.splitext(label_name)[0] + '.txt'))
            
            pbar.update()

ImportError: cannot import name ‘IS_MLU_AVAILABLE’ from ‘mmcv.utils’ (unknown location)

然而。。。

解决

直接扒源码了，mmcv.utils中直接新建一个device_type.py

# Copyright (c) OpenMMLab. All rights reserved.


def is_ipu_available() -> bool:
    try:
        import poptorch
        return poptorch.ipuHardwareIsAvailable()
    except ImportError:
        return False


IS_IPU_AVAILABLE = is_ipu_available()


def is_mlu_available() -> bool:
    try:
        import torch
        return (hasattr(torch, 'is_mlu_available')
                and torch.is_mlu_available())
    except Exception:
        return False


IS_MLU_AVAILABLE = is_mlu_available()


def is_mps_available() -> bool:
    """Return True if mps devices exist.

    It's specialized for mac m1 chips and require torch version 1.12 or higher.
    """
    try:
        import torch
        return hasattr(torch.backends,
                       'mps') and torch.backends.mps.is_available()
    except Exception:
        return False


IS_MPS_AVAILABLE = is_mps_available()


def is_npu_available() -> bool:
    """Return True if npu devices exist."""
    try:
        import torch
        import torch_npu
        return (hasattr(torch, 'npu') and torch_npu.npu.is_available())
    except Exception:
        return False


IS_NPU_AVAILABLE = is_npu_available()

ImportError: cannot import name ‘get_git_hash’ from ‘mmcv.utils’

查阅了FAQ文档后发现，get_git_hash位于mmcv.utils.version_utils

解决

from mmcv.utils.version_utils import get_git_hash

ImportError: cannot import name ‘print_log’ from ‘mmcv’ (unknown location)

解决

from mmcv.utils.logging import print_log

ImportError: cannot import name ‘TORCH_VERSION’ from ‘mmcv.utils’ (unknown location)

解决

from mmcv.utils.parrots_wrapper import TORCH_VERSION

其余以此为参考查阅FAQ文档即可

上述也可直接在utils中新建个init文件即可

ImportError: cannot import name ‘_InstanceNorm’ ，‘_BatchNorm’ from mmcv.utils

在FAQ文档中搜索无果

解决

from torch.nn.modules.instancenorm import _InstanceNorm
from torch.nn.modules.batchnorm import _BatchNorm

ModuleNotFoundError: No module named ‘mmcv._ext’

解决

这个问题简直绝了，前仆后继的遇到，，最终我也是把mmcv-full的版本改为了1.4.0