Fix type annotations for device type

Summary: Fix type annotations for device type

Reviewed By: nikhilaravi

Differential Revision: D28971179

fbshipit-source-id: 410b673c76dfd65ac51b2d144f17ed86a04a3058

pytorch3d/datasets/r2n2/r2n2.py

@@ -9,6 +9,7 @@ from typing import Dict, List, Optional
 import numpy as np
 import torch
 from PIL import Image
+from pytorch3d.common.types import Device
 from pytorch3d.datasets.shapenet_base import ShapeNetBase
 from pytorch3d.renderer import HardPhongShader
 from tabulate import tabulate
@@ -371,7 +372,7 @@ class R2N2(ShapeNetBase):
         idxs: Optional[List[int]] = None,
         view_idxs: Optional[List[int]] = None,
         shader_type=HardPhongShader,
-        device="cpu",
+        device: Device = "cpu",
         **kwargs
     ) -> torch.Tensor:
         """
@@ -394,7 +395,7 @@ class R2N2(ShapeNetBase):
             idxs: List[int] of indices of models to be rendered in the dataset.
             shader_type: Shader to use for rendering. Examples include HardPhongShader
             (default), SoftPhongShader etc or any other type of valid Shader class.
-            device: torch.device on which the tensors should be located.
+            device: Device (as str or torch.device) on which the tensors should be located.
             **kwargs: Accepts any of the kwargs that the renderer supports and any of the
                 args that BlenderCamera supports.
 

pytorch3d/datasets/r2n2/utils.py

@@ -4,6 +4,7 @@ from typing import Dict, List
 
 import numpy as np
 import torch
+from pytorch3d.common.types import Device
 from pytorch3d.datasets.utils import collate_batched_meshes
 from pytorch3d.ops import cubify
 from pytorch3d.renderer import (
@@ -431,13 +432,13 @@ class BlenderCamera(CamerasBase):
     (which uses Blender for rendering the views for each model).
     """
 
-    def __init__(self, R=r, T=t, K=k, device="cpu"):
+    def __init__(self, R=r, T=t, K=k, device: Device = "cpu"):
         """
         Args:
             R: Rotation matrix of shape (N, 3, 3).
             T: Translation matrix of shape (N, 3).
             K: Intrinsic matrix of shape (N, 4, 4).
-            device: torch.device or str.
+            device: Device (as str or torch.device).
         """
         # The initializer formats all inputs to torch tensors and broadcasts
         # all the inputs to have the same batch dimension where necessary.
@@ -450,7 +451,7 @@ class BlenderCamera(CamerasBase):
 
 
 def render_cubified_voxels(
-    voxels: torch.Tensor, shader_type=HardPhongShader, device="cpu", **kwargs
+    voxels: torch.Tensor, shader_type=HardPhongShader, device: Device = "cpu", **kwargs
 ):
     """
     Use the Cubify operator to convert inputs voxels to a mesh and then render that mesh.
@@ -461,7 +462,7 @@ def render_cubified_voxels(
         shader_type: shader_type: shader_type: Shader to use for rendering. Examples
             include HardPhongShader (default), SoftPhongShader etc or any other type
             of valid Shader class.
-        device: torch.device on which the tensors should be located.
+        device: Device (as str or torch.device) on which the tensors should be located.
         **kwargs: Accepts any of the kwargs that the renderer supports.
     Returns:
         Batch of rendered images of shape (N, H, W, 3).

pytorch3d/datasets/shapenet_base.py

@@ -4,6 +4,7 @@ import warnings
 from typing import Dict, List, Optional, Tuple
 
 import torch
+from pytorch3d.common.types import Device
 from pytorch3d.io import load_obj
 from pytorch3d.renderer import (
     FoVPerspectiveCameras,
@@ -105,7 +106,7 @@ class ShapeNetBase(torch.utils.data.Dataset):
         sample_nums: Optional[List[int]] = None,
         idxs: Optional[List[int]] = None,
         shader_type=HardPhongShader,
-        device="cpu",
+        device: Device = "cpu",
         **kwargs
     ) -> torch.Tensor:
         """
@@ -129,7 +130,7 @@ class ShapeNetBase(torch.utils.data.Dataset):
             shader_type: Select shading. Valid options include HardPhongShader (default),
                 SoftPhongShader, HardGouraudShader, SoftGouraudShader, HardFlatShader,
                 SoftSilhouetteShader.
-            device: torch.device on which the tensors should be located.
+            device: Device (as str or torch.device) on which the tensors should be located.
             **kwargs: Accepts any of the kwargs that the renderer supports.
 
         Returns:

pytorch3d/io/mtl_io.py

@@ -9,6 +9,7 @@ import numpy as np
 import torch
 import torch.nn.functional as F
 from iopath.common.file_io import PathManager
+from pytorch3d.common.types import Device
 from pytorch3d.io.utils import _open_file, _read_image
 
 
@@ -393,7 +394,7 @@ TextureImages = Dict[str, torch.Tensor]
 
 
 def _parse_mtl(
-    f, path_manager: PathManager, device="cpu"
+    f, path_manager: PathManager, device: Device = "cpu"
 ) -> Tuple[MaterialProperties, TextureFiles]:
     material_properties = {}
     texture_files = {}
@@ -474,7 +475,7 @@ def load_mtl(
     *,
     material_names: List[str],
     data_dir: str,
-    device="cpu",
+    device: Device = "cpu",
     path_manager: PathManager,
 ) -> Tuple[MaterialProperties, TextureImages]:
     """
@@ -485,6 +486,7 @@ def load_mtl(
         f: a file-like object of the material information.
         material_names: a list of the material names found in the .obj file.
         data_dir: the directory where the material texture files are located.
+        device: Device (as str or torch.tensor) on which to return the new tensors.
         path_manager: PathManager for interpreting both f and material_names.
 
     Returns:

pytorch3d/io/obj_io.py

@@ -11,6 +11,7 @@ from typing import List, Optional, Union
 import numpy as np
 import torch
 from iopath.common.file_io import PathManager
+from pytorch3d.common.types import Device
 from pytorch3d.io.mtl_io import load_mtl, make_mesh_texture_atlas
 from pytorch3d.io.utils import _check_faces_indices, _make_tensor, _open_file
 from pytorch3d.renderer import TexturesAtlas, TexturesUV
@@ -71,7 +72,7 @@ def load_obj(
     create_texture_atlas: bool = False,
     texture_atlas_size: int = 4,
     texture_wrap: Optional[str] = "repeat",
-    device="cpu",
+    device: Device = "cpu",
     path_manager: Optional[PathManager] = None,
 ):
     """
@@ -143,7 +144,7 @@ def load_obj(
             is ignored and a repeating pattern is formed.
             If `texture_mode="clamp"` the values are clamped to the range [0, 1].
             If None, then there is no transformation of the texture values.
-        device: string or torch.device on which to return the new tensors.
+        device: Device (as str or torch.device) on which to return the new tensors.
         path_manager: optionally a PathManager object to interpret paths.
 
     Returns:
@@ -226,7 +227,7 @@ def load_obj(
 
 def load_objs_as_meshes(
     files: list,
-    device=None,
+    device: Optional[Device] = None,
     load_textures: bool = True,
     create_texture_atlas: bool = False,
     texture_atlas_size: int = 4,
@@ -293,7 +294,7 @@ class MeshObjFormat(MeshFormatInterpreter):
         self,
         path: Union[str, Path],
         include_textures: bool,
-        device,
+        device: Device,
         path_manager: PathManager,
         create_texture_atlas: bool = False,
         texture_atlas_size: int = 4,
@@ -497,7 +498,7 @@ def _load_materials(
     *,
     data_dir: str,
     load_textures: bool,
-    device,
+    device: Device,
     path_manager: PathManager,
 ):
     """
@@ -508,7 +509,7 @@ def _load_materials(
         f: a file-like object of the material information.
         data_dir: the directory where the material texture files are located.
         load_textures: whether textures should be loaded.
-        device: string or torch.device on which to return the new tensors.
+        device: Device (as str or torch.device) on which to return the new tensors.
         path_manager: PathManager object to interpret paths.
 
     Returns:
@@ -546,7 +547,7 @@ def _load_obj(
     texture_atlas_size: int = 4,
     texture_wrap: Optional[str] = "repeat",
     path_manager: PathManager,
-    device="cpu",
+    device: Device = "cpu",
 ):
     """
     Load a mesh from a file-like object. See load_obj function more details.

pytorch3d/io/pluggable.py

@@ -8,6 +8,7 @@ from pathlib import Path
 from typing import Deque, Optional, Union
 
 from iopath.common.file_io import PathManager
+from pytorch3d.common.types import Device
 from pytorch3d.structures import Meshes, Pointclouds
 
 from .obj_io import MeshObjFormat
@@ -108,7 +109,7 @@ class IO:
         self,
         path: Union[str, Path],
         include_textures: bool = True,
-        device="cpu",
+        device: Device = "cpu",
         **kwargs,
     ) -> Meshes:
         """
@@ -168,14 +169,14 @@ class IO:
         raise ValueError(f"No mesh interpreter found to write to {path}.")
 
     def load_pointcloud(
-        self, path: Union[str, Path], device="cpu", **kwargs
+        self, path: Union[str, Path], device: Device = "cpu", **kwargs
     ) -> Pointclouds:
         """
         Attempt to load a point cloud from the given file, using a registered format.
 
         Args:
             path: file to read
-            device: torch.device on which to load the data.
+            device: Device (as str or torch.device) on which to load the data.
 
         Returns:
             new Pointclouds object containing one mesh.

pytorch3d/io/utils.py

@@ -10,6 +10,8 @@ import torch
 from iopath.common.file_io import PathManager
 from PIL import Image
 
+from ..common.types import Device
+
 
 @contextlib.contextmanager
 def nullcontext(x):
@@ -31,7 +33,7 @@ def _open_file(f, path_manager: PathManager, mode="r") -> ContextManager[IO]:
 
 
 def _make_tensor(
-    data, cols: int, dtype: torch.dtype, device: str = "cpu"
+    data, cols: int, dtype: torch.dtype, device: Device = "cpu"
 ) -> torch.Tensor:
     """
     Return a 2D tensor with the specified cols and dtype filled with data,

pytorch3d/renderer/cameras.py

@@ -7,6 +7,7 @@ from typing import Optional, Sequence, Tuple
 import numpy as np
 import torch
 import torch.nn.functional as F
+from pytorch3d.common.types import Device
 from pytorch3d.transforms import Rotate, Transform3d, Translate
 
 from .utils import TensorProperties, convert_to_tensors_and_broadcast
@@ -290,7 +291,7 @@ def OpenGLPerspectiveCameras(
     degrees: bool = True,
     R=_R,
     T=_T,
-    device="cpu",
+    device: Device = "cpu",
 ):
     """
     OpenGLPerspectiveCameras has been DEPRECATED. Use FoVPerspectiveCameras instead.
@@ -358,7 +359,7 @@ class FoVPerspectiveCameras(CamerasBase):
         R=_R,
         T=_T,
         K=None,
-        device="cpu",
+        device: Device = "cpu",
     ):
         """
 
@@ -373,7 +374,7 @@ class FoVPerspectiveCameras(CamerasBase):
             T: Translation matrix of shape (N, 3)
             K: (optional) A calibration matrix of shape (N, 4, 4)
                 If provided, don't need znear, zfar, fov, aspect_ratio, degrees
-            device: torch.device or string
+            device: Device (as str or torch.device)
         """
         # The initializer formats all inputs to torch tensors and broadcasts
         # all the inputs to have the same batch dimension where necessary.

pytorch3d/renderer/lighting.py

@@ -4,6 +4,7 @@
 import torch
 import torch.nn.functional as F
 
+from ..common.types import Device
 from .utils import TensorProperties, convert_to_tensors_and_broadcast
 
 
@@ -158,7 +159,7 @@ class DirectionalLights(TensorProperties):
         diffuse_color=((0.3, 0.3, 0.3),),
         specular_color=((0.2, 0.2, 0.2),),
         direction=((0, 1, 0),),
-        device="cpu",
+        device: Device = "cpu",
     ):
         """
         Args:
@@ -166,7 +167,7 @@ class DirectionalLights(TensorProperties):
             diffuse_color: RGB color of the diffuse component.
             specular_color: RGB color of the specular component.
             direction: (x, y, z) direction vector of the light.
-            device: torch.device on which the tensors should be located
+            device: Device (as str or torch.device) on which the tensors should be located
 
         The inputs can each be
             - 3 element tuple/list or list of lists
@@ -219,7 +220,7 @@ class PointLights(TensorProperties):
         diffuse_color=((0.3, 0.3, 0.3),),
         specular_color=((0.2, 0.2, 0.2),),
         location=((0, 1, 0),),
-        device="cpu",
+        device: Device = "cpu",
     ):
         """
         Args:
@@ -227,7 +228,7 @@ class PointLights(TensorProperties):
             diffuse_color: RGB color of the diffuse component
             specular_color: RGB color of the specular component
             location: xyz position of the light.
-            device: torch.device on which the tensors should be located
+            device: Device (as str or torch.device) on which the tensors should be located
 
         The inputs can each be
             - 3 element tuple/list or list of lists
@@ -275,14 +276,14 @@ class AmbientLights(TensorProperties):
     not used in rendering.
     """
 
-    def __init__(self, *, ambient_color=None, device="cpu"):
+    def __init__(self, *, ambient_color=None, device: Device = "cpu"):
         """
         If ambient_color is provided, it should be a sequence of
         triples of floats.
 
         Args:
             ambient_color: RGB color
-            device: torch.device on which the tensors should be located
+            device: Device (as str or torch.device) on which the tensors should be located
 
         The ambient_color if provided, should be
             - 3 element tuple/list or list of lists

pytorch3d/renderer/materials.py

@@ -3,6 +3,7 @@
 
 import torch
 
+from ..common.types import Device
 from .utils import TensorProperties
 
 
@@ -18,7 +19,7 @@ class Materials(TensorProperties):
         diffuse_color=((1, 1, 1),),
         specular_color=((1, 1, 1),),
         shininess=64,
-        device="cpu",
+        device: Device = "cpu",
     ):
         """
         Args:
@@ -29,7 +30,7 @@ class Materials(TensorProperties):
                 the focus of the specular highlight with a high value
                 resulting in a concentrated highlight. Shininess values
                 can range from 0-1000.
-            device: torch.device or string
+            device: Device (as str or torch.device) on which the tensors should be located
 
         ambient_color, diffuse_color and specular_color can be of shape
         (1, 3) or (N, 3). shininess can be of shape (1) or (N).

pytorch3d/renderer/mesh/shader.py

@@ -5,6 +5,7 @@ import warnings
 import torch
 import torch.nn as nn
 
+from ...common.types import Device
 from ..blending import (
     BlendParams,
     hard_rgb_blend,
@@ -40,7 +41,12 @@ class HardPhongShader(nn.Module):
     """
 
     def __init__(
-        self, device="cpu", cameras=None, lights=None, materials=None, blend_params=None
+        self,
+        device: Device = "cpu",
+        cameras=None,
+        lights=None,
+        materials=None,
+        blend_params=None,
     ):
         super().__init__()
         self.lights = lights if lights is not None else PointLights(device=device)
@@ -50,7 +56,7 @@ class HardPhongShader(nn.Module):
         self.cameras = cameras
         self.blend_params = blend_params if blend_params is not None else BlendParams()
 
-    def to(self, device):
+    def to(self, device: Device):
         # Manually move to device modules which are not subclasses of nn.Module
         self.cameras = self.cameras.to(device)
         self.materials = self.materials.to(device)
@@ -95,7 +101,12 @@ class SoftPhongShader(nn.Module):
     """
 
     def __init__(
-        self, device="cpu", cameras=None, lights=None, materials=None, blend_params=None
+        self,
+        device: Device = "cpu",
+        cameras=None,
+        lights=None,
+        materials=None,
+        blend_params=None,
     ):
         super().__init__()
         self.lights = lights if lights is not None else PointLights(device=device)
@@ -105,7 +116,7 @@ class SoftPhongShader(nn.Module):
         self.cameras = cameras
         self.blend_params = blend_params if blend_params is not None else BlendParams()
 
-    def to(self, device):
+    def to(self, device: Device):
         # Manually move to device modules which are not subclasses of nn.Module
         self.cameras = self.cameras.to(device)
         self.materials = self.materials.to(device)
@@ -155,7 +166,12 @@ class HardGouraudShader(nn.Module):
     """
 
     def __init__(
-        self, device="cpu", cameras=None, lights=None, materials=None, blend_params=None
+        self,
+        device: Device = "cpu",
+        cameras=None,
+        lights=None,
+        materials=None,
+        blend_params=None,
     ):
         super().__init__()
         self.lights = lights if lights is not None else PointLights(device=device)
@@ -165,7 +181,7 @@ class HardGouraudShader(nn.Module):
         self.cameras = cameras
         self.blend_params = blend_params if blend_params is not None else BlendParams()
 
-    def to(self, device):
+    def to(self, device: Device):
         # Manually move to device modules which are not subclasses of nn.Module
         self.cameras = self.cameras.to(device)
         self.materials = self.materials.to(device)
@@ -214,7 +230,12 @@ class SoftGouraudShader(nn.Module):
     """
 
     def __init__(
-        self, device="cpu", cameras=None, lights=None, materials=None, blend_params=None
+        self,
+        device: Device = "cpu",
+        cameras=None,
+        lights=None,
+        materials=None,
+        blend_params=None,
     ):
         super().__init__()
         self.lights = lights if lights is not None else PointLights(device=device)
@@ -224,7 +245,7 @@ class SoftGouraudShader(nn.Module):
         self.cameras = cameras
         self.blend_params = blend_params if blend_params is not None else BlendParams()
 
-    def to(self, device):
+    def to(self, device: Device):
         # Manually move to device modules which are not subclasses of nn.Module
         self.cameras = self.cameras.to(device)
         self.materials = self.materials.to(device)
@@ -255,7 +276,7 @@ class SoftGouraudShader(nn.Module):
 
 
 def TexturedSoftPhongShader(
-    device="cpu", cameras=None, lights=None, materials=None, blend_params=None
+    device: Device = "cpu", cameras=None, lights=None, materials=None, blend_params=None
 ):
     """
     TexturedSoftPhongShader class has been DEPRECATED. Use SoftPhongShader instead.
@@ -290,7 +311,12 @@ class HardFlatShader(nn.Module):
     """
 
     def __init__(
-        self, device="cpu", cameras=None, lights=None, materials=None, blend_params=None
+        self,
+        device: Device = "cpu",
+        cameras=None,
+        lights=None,
+        materials=None,
+        blend_params=None,
     ):
         super().__init__()
         self.lights = lights if lights is not None else PointLights(device=device)
@@ -300,7 +326,7 @@ class HardFlatShader(nn.Module):
         self.cameras = cameras
         self.blend_params = blend_params if blend_params is not None else BlendParams()
 
-    def to(self, device):
+    def to(self, device: Device):
         # Manually move to device modules which are not subclasses of nn.Module
         self.cameras = self.cameras.to(device)
         self.materials = self.materials.to(device)

pytorch3d/renderer/utils.py

@@ -10,6 +10,8 @@ import numpy as np
 import torch
 import torch.nn as nn
 
+from ..common.types import Device, make_device
+
 
 class TensorAccessor(nn.Module):
     """
@@ -88,17 +90,19 @@ class TensorProperties(nn.Module):
     A mix-in class for storing tensors as properties with helper methods.
     """
 
-    def __init__(self, dtype=torch.float32, device="cpu", **kwargs):
+    def __init__(
+        self, dtype: torch.dtype = torch.float32, device: Device = "cpu", **kwargs
+    ):
         """
         Args:
             dtype: data type to set for the inputs
-            device: str or torch.device
+            device: Device (as str or torch.device)
             kwargs: any number of keyword arguments. Any arguments which are
-                of type (float/int/tuple/tensor/array) are broadcasted and
+                of type (float/int/list/tuple/tensor/array) are broadcasted and
                 other keyword arguments are set as attributes.
         """
         super().__init__()
-        self.device = device
+        self.device = make_device(device)
         self._N = 0
         if kwargs is not None:
 
@@ -108,7 +112,7 @@ class TensorProperties(nn.Module):
             for k, v in kwargs.items():
                 if v is None or isinstance(v, (str, bool)):
                     setattr(self, k, v)
-                elif isinstance(v, BROADCAST_TYPES):
+                elif isinstance(v, BROADCAST_TYPES):  # pyre-fixme[6]
                     args_to_broadcast[k] = v
                 else:
                     msg = "Arg %s with type %r is not broadcastable"
@@ -152,17 +156,18 @@ class TensorProperties(nn.Module):
         msg = "Expected index of type int or slice; got %r"
         raise ValueError(msg % type(index))
 
-    def to(self, device: str = "cpu"):
+    def to(self, device: Device = "cpu"):
         """
         In place operation to move class properties which are tensors to a
         specified device. If self has a property "device", update this as well.
         """
+        device_ = make_device(device)
         for k in dir(self):
             v = getattr(self, k)
             if k == "device":
-                setattr(self, k, device)
-            if torch.is_tensor(v) and v.device != device:
-                setattr(self, k, v.to(device))
+                setattr(self, k, device_)
+            if torch.is_tensor(v) and v.device != device_:
+                setattr(self, k, v.to(device_))
         return self
 
     def clone(self, other):
@@ -257,28 +262,37 @@ class TensorProperties(nn.Module):
         return self
 
 
-def format_tensor(input, dtype=torch.float32, device: str = "cpu") -> torch.Tensor:
+def format_tensor(
+    input, dtype: torch.dtype = torch.float32, device: Device = "cpu"
+) -> torch.Tensor:
     """
     Helper function for converting a scalar value to a tensor.
 
     Args:
         input: Python scalar, Python list/tuple, torch scalar, 1D torch tensor
         dtype: data type for the input
-        device: torch device on which the tensor should be placed.
+        device: Device (as str or torch.device) on which the tensor should be placed.
 
     Returns:
         input_vec: torch tensor with optional added batch dimension.
     """
+    device_ = make_device(device)
     if not torch.is_tensor(input):
-        input = torch.tensor(input, dtype=dtype, device=device)
+        input = torch.tensor(input, dtype=dtype, device=device_)
+
     if input.dim() == 0:
         input = input.view(1)
-    if input.device != device:
-        input = input.to(device=device)
+
+    if input.device == device_:
+        return input
+
+    input = input.to(device=device)
     return input
 
 
-def convert_to_tensors_and_broadcast(*args, dtype=torch.float32, device: str = "cpu"):
+def convert_to_tensors_and_broadcast(
+    *args, dtype: torch.dtype = torch.float32, device: Device = "cpu"
+):
     """
     Helper function to handle parsing an arbitrary number of inputs (*args)
     which all need to have the same batch dimension.