[Enhance] Remove unnecessary calls and lazily import to speed import performance (#837)

* [Enhance] Remove unnecessary calls to speed import performance

* lazily import matplotlib

* minor refinement

mmengine/runner/checkpoint.py

@@ -12,7 +12,6 @@ from tempfile import TemporaryDirectory
 from typing import Callable, Dict, Optional
 
 import torch
-import torchvision
 
 import mmengine
 from mmengine.dist import get_dist_info
@@ -112,6 +111,7 @@ def load_state_dict(module, state_dict, strict=False, logger=None):
 
 
 def get_torchvision_models():
+    import torchvision
     if digit_version(torchvision.__version__) < digit_version('0.13.0a0'):
         model_urls = dict()
         # When the version of torchvision is lower than 0.13, the model url is

mmengine/utils/dl_utils/collect_env.py

@@ -4,7 +4,6 @@ import os.path as osp
 import subprocess
 import sys
 from collections import OrderedDict, defaultdict
-from distutils import errors
 
 import cv2
 import numpy as np
@@ -47,6 +46,8 @@ def collect_env():
             - OpenCV (optional): OpenCV version.
             - MMENGINE: MMENGINE version.
     """
+    from distutils import errors
+
     env_info = OrderedDict()
     env_info['sys.platform'] = sys.platform
     env_info['Python'] = sys.version.replace('\n', '')

mmengine/utils/dl_utils/parrots_wrapper.py

@@ -103,7 +103,6 @@ def _get_norm() -> tuple:
 
 _ConvNd, _ConvTransposeMixin = _get_conv()
 DataLoader, PoolDataLoader = _get_dataloader()
-BuildExtension, CppExtension, CUDAExtension = _get_extension()
 _BatchNorm, _InstanceNorm, SyncBatchNorm_ = _get_norm()
 _AdaptiveAvgPoolNd, _AdaptiveMaxPoolNd, _AvgPoolNd, _MaxPoolNd = _get_pool()
 

mmengine/utils/package_utils.py

@@ -3,9 +3,6 @@ import importlib
 import os.path as osp
 import subprocess
 
-import pkg_resources
-from pkg_resources import get_distribution
-
 
 def is_installed(package: str) -> bool:
     """Check package whether installed.
@@ -13,6 +10,12 @@ def is_installed(package: str) -> bool:
     Args:
         package (str): Name of package to be checked.
     """
+    # When executing `import mmengine.runner`,
+    # pkg_resources will be imported and it takes too much time.
+    # Therefore, import it in function scope to save time.
+    import pkg_resources
+    from pkg_resources import get_distribution
+
     # refresh the pkg_resources
     # more datails at https://github.com/pypa/setuptools/issues/373
     importlib.reload(pkg_resources)
@@ -33,6 +36,8 @@ def get_installed_path(package: str) -> str:
         >>> get_installed_path('mmcls')
         >>> '.../lib/python3.7/site-packages/mmcls'
     """
+    from pkg_resources import get_distribution
+
     # if the package name is not the same as module name, module name should be
     # inferred. For example, mmcv-full is the package name, but mmcv is module
     # name. If we want to get the installed path of mmcv-full, we should concat
@@ -51,6 +56,7 @@ def package2module(package: str):
     Args:
         package (str): Package to infer module name.
     """
+    from pkg_resources import get_distribution
     pkg = get_distribution(package)
     if pkg.has_metadata('top_level.txt'):
         module_name = pkg.get_metadata('top_level.txt').split('\n')[0]

mmengine/visualization/utils.py

 # Copyright (c) OpenMMLab. All rights reserved.
 
-from typing import Any, List, Optional, Tuple, Type, Union
+from typing import TYPE_CHECKING, Any, List, Optional, Tuple, Type, Union
 
 import cv2
-import matplotlib
-import matplotlib.pyplot as plt
 import numpy as np
 import torch
-from matplotlib.backend_bases import CloseEvent
-from matplotlib.backends.backend_agg import FigureCanvasAgg
+
+if TYPE_CHECKING:
+    from matplotlib.backends.backend_agg import FigureCanvasAgg
 
 
 def tensor2ndarray(value: Union[np.ndarray, torch.Tensor]) -> np.ndarray:
@@ -131,6 +130,7 @@ def color_str2rgb(color: str) -> tuple:
     Returns:
         tuple: RGB color.
     """
+    import matplotlib
     rgb_color: tuple = matplotlib.colors.to_rgb(color)
     rgb_color = tuple(int(c * 255) for c in rgb_color)
     return rgb_color
@@ -186,6 +186,8 @@ def wait_continue(figure, timeout: int = 0, continue_key: str = ' ') -> int:
         int: If zero, means time out or the user pressed ``continue_key``,
             and if one, means the user closed the show figure.
     """  # noqa: E501
+    import matplotlib.pyplot as plt
+    from matplotlib.backend_bases import CloseEvent
     is_inline = 'inline' in plt.get_backend()
     if is_inline:
         # If use inline backend, interactive input and timeout is no use.
@@ -226,7 +228,7 @@ def wait_continue(figure, timeout: int = 0, continue_key: str = ' ') -> int:
             return 0  # Quit for continue.
 
 
-def img_from_canvas(canvas: FigureCanvasAgg) -> np.ndarray:
+def img_from_canvas(canvas: 'FigureCanvasAgg') -> np.ndarray:
     """Get RGB image from ``FigureCanvasAgg``.
 
     Args:

mmengine/visualization/visualizer.py

@@ -4,16 +4,9 @@ import warnings
 from typing import Dict, List, Optional, Sequence, Tuple, Union
 
 import cv2
-import matplotlib.pyplot as plt
 import numpy as np
 import torch
 import torch.nn.functional as F
-from matplotlib.backends.backend_agg import FigureCanvasAgg
-from matplotlib.collections import (LineCollection, PatchCollection,
-                                    PolyCollection)
-from matplotlib.figure import Figure
-from matplotlib.patches import Circle
-from matplotlib.pyplot import new_figure_manager
 
 from mmengine.config import Config
 from mmengine.dist import master_only
@@ -240,6 +233,7 @@ class Visualizer(ManagerMixin):
             continue_key (str): The key for users to continue. Defaults to
                 the space key.
         """
+        import matplotlib.pyplot as plt
         is_inline = 'inline' in plt.get_backend()
         img = self.get_image() if drawn_img is None else drawn_img
         self._init_manager(win_name)
@@ -302,7 +296,8 @@ class Visualizer(ManagerMixin):
         Returns:
              tuple: build canvas figure and axes.
         """
-
+        from matplotlib.backends.backend_agg import FigureCanvasAgg
+        from matplotlib.figure import Figure
         fig = Figure(**fig_cfg)
         ax = fig.add_subplot()
         ax.axis(False)
@@ -318,6 +313,8 @@ class Visualizer(ManagerMixin):
         Args:
             win_name (str): The window name.
         """
+        from matplotlib.figure import Figure
+        from matplotlib.pyplot import new_figure_manager
         if getattr(self, 'manager', None) is None:
             self.manager = new_figure_manager(
                 num=1, FigureClass=Figure, **self.fig_show_cfg)
@@ -546,6 +543,7 @@ class Visualizer(ManagerMixin):
                 If ``line_widths`` is single value, all the lines will
                 have the same linewidth. Defaults to 2.
         """
+        from matplotlib.collections import LineCollection
         check_type('x_datas', x_datas, (np.ndarray, torch.Tensor))
         x_datas = tensor2ndarray(x_datas)
         check_type('y_datas', y_datas, (np.ndarray, torch.Tensor))
@@ -614,6 +612,8 @@ class Visualizer(ManagerMixin):
             alpha (Union[int, float]): The transparency of circles.
                 Defaults to 0.8.
         """
+        from matplotlib.collections import PatchCollection
+        from matplotlib.patches import Circle
         check_type('center', center, (np.ndarray, torch.Tensor))
         center = tensor2ndarray(center)
         check_type('radius', radius, (np.ndarray, torch.Tensor))
@@ -760,6 +760,7 @@ class Visualizer(ManagerMixin):
             alpha (Union[int, float]): The transparency of polygons.
                 Defaults to 0.8.
         """
+        from matplotlib.collections import PolyCollection
         check_type('polygons', polygons, (list, np.ndarray, torch.Tensor))
         edge_colors = color_val_matplotlib(edge_colors)  # type: ignore
         face_colors = color_val_matplotlib(face_colors)  # type: ignore
@@ -916,6 +917,7 @@ class Visualizer(ManagerMixin):
         Returns:
             np.ndarray: RGB image.
         """
+        import matplotlib.pyplot as plt
         assert isinstance(featmap,
                           torch.Tensor), (f'`featmap` should be torch.Tensor,'
                                           f' but got {type(featmap)}')