Final typing fixes + add Pyright to CI (#1944)

* knocking out a couple

* Fix more errors

* Add pyright

* Remove some unnecessary cv2 type ignores, warning on missing imports
(for missing tinycudann)

* Minor annotation fixes

* Update docs

* Drop `None` from temporal distortion output

---------

Co-authored-by: Brent Yi <yibrenth@gmail.com>
This commit is contained in:
Matthew TancikandBrent Yi authored and GitHub committed 2023-05-18 02:25:34 -07:00
1 parent 08e3a1f1a6
commit 2b0831b6d4
28 files changed
+110 -71

No files matched your search

+3
View File
@@ -36,6 +36,9 @@ jobs:
run: ruff docs/ nerfstudio/ tests/
- name: Run Black
run: black docs/ nerfstudio/ tests/ --check
- name: Run Pyright
run: |
pyright
- name: Test with pytest
run: |
pytest
+3 -1
View File
@@ -18,6 +18,7 @@ Below are the various tooling features our team uses to maintain this codebase.
| --------------- | ---------------------------------------------------------- |
| Formatting | [Black](https://black.readthedocs.io/en/stable/) |
| Linter | [Ruff](https://beta.ruff.rs/docs/) |
| Type checking | [Pyright](https://github.com/microsoft/pyright) |
| Testing | [pytest](https://docs.pytest.org/en/7.1.x/) |
| Docs | [Sphinx](https://www.sphinx-doc.org/en/master/) |
| Docstring style | [Google](https://google.github.io/styleguide/pyguide.html) |
@@ -53,7 +54,8 @@ conda install -c conda-forge pandoc
This will perform the following checks and actions:
- Black/ Linting style check: Ensures code is consistently and properly formatted.
- Formatting and linting: Ensures code is consistently and properly formatted.
- Type checking: Ensures static type safety.
- Pytests: Runs pytests locally to make sure added code does not break existing logic.
- Documentation build: Builds docs locally. Ensures changes do not result in warnings/errors.
- Licensing: Automatically adds licensing headers to the correct files.
@@ -38,7 +38,6 @@ from typing import (
import torch
from torch import nn
from torch.nn import Parameter
from torch.utils.data import Dataset
from torch.utils.data.distributed import DistributedSampler
from typing_extensions import TypeVar
@@ -154,8 +153,8 @@ class DataManager(nn.Module):
"""
train_dataset: Optional[Dataset] = None
eval_dataset: Optional[Dataset] = None
train_dataset: Optional[InputDataset] = None
eval_dataset: Optional[InputDataset] = None
train_sampler: Optional[DistributedSampler] = None
eval_sampler: Optional[DistributedSampler] = None
includes_time: bool = False
@@ -278,9 +277,9 @@ class DataManager(nn.Module):
"""Returns the number of rays per batch for evaluation."""
raise NotImplementedError
def get_datapath(self) -> Optional[Path]:
@abstractmethod
def get_datapath(self) -> Path:
"""Returns the path to the data. This is used to determine where to save camera paths."""
return None
def get_training_callbacks(
self, training_callback_attributes: TrainingCallbackAttributes
@@ -325,7 +324,7 @@ class VanillaDataManagerConfig(DataManagerConfig):
camera_optimizer: CameraOptimizerConfig = CameraOptimizerConfig()
"""Specifies the camera pose optimizer used during training. Helpful if poses are noisy, such as for data from
Record3D."""
collate_fn: Callable[[Any], Any] = staticmethod(nerfstudio_collate)
collate_fn: Callable[[Any], Any] = nerfstudio_collate
"""Specifies the collate function to use for the train and eval dataloaders."""
camera_res_scale_factor: float = 1.0
"""The scale factor for scaling spatial data such as images, mask, semantics
@@ -74,7 +74,7 @@ def upscale(img, scale: int) -> np.ndarray:
return resized
def rescale(img, scale_factor: float, interpolation=cv2.INTER_AREA) -> np.ndarray: # type: ignore
def rescale(img, scale_factor: float, interpolation: int = cv2.INTER_AREA) -> np.ndarray:
"""Function from DyCheck's repo. Rescale an image.
Args:
@@ -316,13 +316,13 @@ class Dycheck(DataParser):
d = self.config.downscale_factor
if not image_filenames[0].exists():
CONSOLE.print(f"downscale factor {d}x not exist, converting")
ori_h, ori_w = cv2.imread(str(self.data / f"rgb/1x/{frame_names[0]}.png")).shape[:2] # type: ignore
ori_h, ori_w = cv2.imread(str(self.data / f"rgb/1x/{frame_names[0]}.png")).shape[:2]
(self.data / f"rgb/{d}x").mkdir(exist_ok=True)
h, w = ori_h // d, ori_w // d
for frame in frame_names:
cv2.imwrite( # type: ignore
cv2.imwrite(
str(self.data / f"rgb/{d}x/{frame}.png"),
cv2.resize(cv2.imread(str(self.data / f"rgb/1x/{frame}.png")), [h, w]), # type: ignore
cv2.resize(cv2.imread(str(self.data / f"rgb/1x/{frame}.png")), (h, w)),
)
CONSOLE.print("finished")
@@ -331,8 +331,8 @@ class Dycheck(DataParser):
(self.data / f"processed_depth/{d}x").mkdir(exist_ok=True, parents=True)
for idx, frame in enumerate(frame_names):
depth = np.load(self.data / f"depth/1x/{frame}.npy")
mask = rescale((depth != 0).astype(np.uint8) * 255, 1 / d, cv2.INTER_AREA) # type: ignore
depth = rescale(depth, 1 / d, cv2.INTER_AREA) # type: ignore
mask = rescale((depth != 0).astype(np.uint8) * 255, 1 / d, cv2.INTER_AREA)
depth = rescale(depth, 1 / d, cv2.INTER_AREA)
depth[mask != 255] = 0
depth = _rescale_depth(depth, cams[idx])
np.save(str(self.data / f"processed_depth/{d}x/{frame}.npy"), depth)
+3
View File
@@ -29,6 +29,7 @@ from PIL import Image
from torch import Tensor
from torch.utils.data import Dataset
from nerfstudio.cameras.cameras import Cameras
from nerfstudio.data.dataparsers.base_dataparser import DataparserOutputs
from nerfstudio.data.utils.data_utils import get_image_mask_tensor_from_path
@@ -41,6 +42,8 @@ class InputDataset(Dataset):
scale_factor: The scaling factor for the dataparser outputs
"""
cameras: Cameras
def __init__(self, dataparser_outputs: DataparserOutputs, scale_factor: float = 1.0):
super().__init__()
self._dataparser_outputs = dataparser_outputs
+4 -4
View File
@@ -61,7 +61,7 @@ def get_depth_image_from_path(
height: int,
width: int,
scale_factor: float,
interpolation: int = cv2.INTER_NEAREST, # type: ignore
interpolation: int = cv2.INTER_NEAREST,
) -> torch.Tensor:
"""Loads, rescales and resizes depth images.
Filepath points to a 16-bit or 32-bit depth image, or a numpy array `*.npy`.
@@ -78,9 +78,9 @@ def get_depth_image_from_path(
"""
if filepath.suffix == ".npy":
image = np.load(filepath) * scale_factor
image = cv2.resize(image, (width, height), interpolation=interpolation) # type: ignore
image = cv2.resize(image, (width, height), interpolation=interpolation)
else:
image = cv2.imread(str(filepath.absolute()), cv2.IMREAD_ANYDEPTH) # type: ignore
image = cv2.imread(str(filepath.absolute()), cv2.IMREAD_ANYDEPTH)
image = image.astype(np.float64) * scale_factor
image = cv2.resize(image, (width, height), interpolation=interpolation) # type: ignore
image = cv2.resize(image, (width, height), interpolation=interpolation)
return torch.from_numpy(image[:, :, np.newaxis])
+2 -2
View File
@@ -21,7 +21,7 @@ import concurrent.futures
import multiprocessing
import random
from abc import abstractmethod
from typing import Dict, Optional, Sized, Tuple, Union
from typing import Any, Callable, Dict, Optional, Sized, Tuple, Union
import torch
from rich.progress import track
@@ -54,7 +54,7 @@ class CacheDataloader(DataLoader):
num_images_to_sample_from: int = -1,
num_times_to_repeat_images: int = -1,
device: Union[torch.device, str] = "cpu",
collate_fn=nerfstudio_collate,
collate_fn: Callable[[Any], Any] = nerfstudio_collate,
**kwargs,
):
self.dataset = dataset
+4 -3
View File
@@ -21,7 +21,7 @@ from __future__ import annotations
import math
from pathlib import Path
from typing import Literal, Tuple, Union
from typing import Literal, Optional, Tuple, Union
import mediapy as media
import numpy as np
@@ -324,7 +324,7 @@ def export_textured_mesh(
mesh: Mesh,
pipeline: Pipeline,
output_dir: Path,
px_per_uv_triangle: int,
px_per_uv_triangle: Optional[int],
unwrap_method: Literal["xatlas", "custom"] = "xatlas",
raylen_method: Literal["edge", "none"] = "edge",
num_pixels_per_side=1024,
@@ -338,7 +338,7 @@ def export_textured_mesh(
mesh: The mesh to texture.
pipeline: The pipeline to use for texturing.
output_dir: The directory to write the textured mesh to.
px_per_uv_triangle: The number of pixels per side of UV triangle.
px_per_uv_triangle: The number of pixels per side of UV triangle. Required for "custom" unwrapping.
unwrap_method: The method to use for unwrapping the mesh.
offset_method: The method to use for computing the ray length to render.
num_pixels_per_side: The number of pixels per side of the texture image.
@@ -363,6 +363,7 @@ def export_textured_mesh(
CONSOLE.print("[bold green]:white_check_mark: Unwrapped mesh with xatlas method")
elif unwrap_method == "custom":
CONSOLE.print("Unwrapping mesh with custom method...")
assert px_per_uv_triangle is not None
texture_coordinates, origins, directions = unwrap_mesh_per_uv_triangle(
vertices, faces, vertex_normals, px_per_uv_triangle
)
+1
View File
@@ -304,6 +304,7 @@ def export_tsdf_mesh(
device = pipeline.device
assert pipeline.datamanager.train_dataset is not None
dataparser_outputs = pipeline.datamanager.train_dataset._dataparser_outputs
# initialize the TSDF volume
@@ -16,7 +16,7 @@
import abc
from enum import Enum
from typing import Any, Dict, Optional, Tuple
from typing import Any, Dict, Tuple
import torch
from jaxtyping import Float
@@ -30,7 +30,7 @@ class TemporalDistortion(nn.Module):
"""Apply spatial distortions as a function of time"""
@abc.abstractmethod
def forward(self, positions: Float[Tensor, "*bs 3"], times: Optional[Float[Tensor, "1"]]) -> Float[Tensor, "*bs 3"]:
def forward(self, positions: Float[Tensor, "*bs 3"], times: Float[Tensor, "*bs 1"]) -> Float[Tensor, "*bs 3"]:
"""
Args:
positions: Samples to translate as a function of time
@@ -86,9 +86,7 @@ class DNeRFDistortion(TemporalDistortion):
skip_connections=skip_connections,
)
def forward(self, positions, times=None):
if times is None:
return None
def forward(self, positions: Float[Tensor, "*bs 3"], times: Float[Tensor, "*bs 1"]) -> Float[Tensor, "*bs 3"]:
p = self.position_encoding(positions)
t = self.temporal_encoding(times)
return self.mlp_deform(torch.cat([p, t], dim=-1))
@@ -33,7 +33,7 @@ class SceneCollider(nn.Module):
self.kwargs = kwargs
super().__init__()
def set_nears_and_fars(self, ray_bundle) -> RayBundle:
def set_nears_and_fars(self, ray_bundle: RayBundle) -> RayBundle:
"""To be implemented."""
raise NotImplementedError
+8 -2
View File
@@ -143,7 +143,11 @@ class NeRFModel(Model):
# uniform sampling
ray_samples_uniform = self.sampler_uniform(ray_bundle)
if self.temporal_distortion is not None:
offsets = self.temporal_distortion(ray_samples_uniform.frustums.get_positions(), ray_samples_uniform.times)
offsets = None
if ray_samples_uniform.times is not None:
offsets = self.temporal_distortion(
ray_samples_uniform.frustums.get_positions(), ray_samples_uniform.times
)
ray_samples_uniform.frustums.set_offsets(offsets)
# coarse field:
@@ -159,7 +163,9 @@ class NeRFModel(Model):
# pdf sampling
ray_samples_pdf = self.sampler_pdf(ray_bundle, ray_samples_uniform, weights_coarse)
if self.temporal_distortion is not None:
offsets = self.temporal_distortion(ray_samples_pdf.frustums.get_positions(), ray_samples_pdf.times)
offsets = None
if ray_samples_pdf.times is not None:
offsets = self.temporal_distortion(ray_samples_pdf.frustums.get_positions(), ray_samples_pdf.times)
ray_samples_pdf.frustums.set_offsets(offsets)
# fine field:
+1 -2
View File
@@ -39,7 +39,6 @@ from torch.nn.parallel import DistributedDataParallel as DDP
from nerfstudio.configs import base_config as cfg
from nerfstudio.data.datamanagers.base_datamanager import (
DataManager,
DataManagerConfig,
VanillaDataManager,
VanillaDataManagerConfig,
)
@@ -205,7 +204,7 @@ class VanillaPipelineConfig(cfg.InstantiateConfig):
_target: Type = field(default_factory=lambda: VanillaPipeline)
"""target class to instantiate"""
datamanager: DataManagerConfig = VanillaDataManagerConfig()
datamanager: VanillaDataManagerConfig = VanillaDataManagerConfig()
"""specifies the datamanager config"""
model: ModelConfig = ModelConfig()
"""specifies the model config"""
@@ -203,6 +203,10 @@ class ColmapConverterToNerfstudioDataset(BaseConverterToNerfstudioDataset):
elif sfm_tool == "hloc":
if mask_path is not None:
raise RuntimeError("Cannot use a mask with hloc. Please remove the cropping options " "and try again.")
assert feature_type is not None
assert matcher_type is not None
assert matcher_type != "NN" # Only used for colmap.
hloc_utils.run_hloc(
image_dir=self.image_dir,
colmap_dir=self.absolute_colmap_path,
+3 -3
View File
@@ -582,7 +582,7 @@ def create_sfm_depth(
out_name = str(im_data.name)
depth_path = output_dir / out_name
cv2.imwrite(str(depth_path), depth_img)
cv2.imwrite(str(depth_path), depth_img) # type: ignore
image_id_to_depth_path[im_id] = depth_path
@@ -594,13 +594,13 @@ def create_sfm_depth(
overlay = 255.0 * colormaps.apply_depth_colormap(torch.from_numpy(depth_flat)).numpy()
overlay = overlay.reshape([H, W, 3])
input_image_path = input_images_dir / im_data.name
input_image = cv2.imread(str(input_image_path))
input_image = cv2.imread(str(input_image_path)) # type: ignore
debug = 0.3 * input_image + 0.7 + overlay
out_name = out_name + ".debug.jpg"
output_path = output_dir / "debug_depth" / out_name
output_path.parent.mkdir(parents=True, exist_ok=True)
cv2.imwrite(str(output_path), debug.astype(np.uint8))
cv2.imwrite(str(output_path), debug.astype(np.uint8)) # type: ignore
return image_id_to_depth_path
@@ -190,6 +190,7 @@ def generate_planar_projections_from_equirectangular(
v_rad = torch.pi * v_deg / 180.0
u_rad = torch.pi * u_deg / 180.0
pers_image = equi2pers(im, rots={"roll": 0, "pitch": v_rad, "yaw": u_rad}) * 255.0
assert isinstance(pers_image, torch.Tensor)
pers_image = (pers_image.permute(1, 2, 0)).type(torch.uint8).to("cpu").numpy()
cv2.imwrite(f"{output_dir}/{i[:-4]}_{count}.jpg", pers_image)
count += 1
+1 -1
View File
@@ -42,7 +42,7 @@ else:
_HAS_HLOC = True
try:
from pixsfm.refine_hloc import PixSfM
from pixsfm.refine_hloc import PixSfM # type: ignore
except ImportError:
_HAS_PIXSFM = False
else:
@@ -395,7 +395,7 @@ def find_tool_feature_matcher_combination(
"sosnet",
"disk",
],
Literal["superglue", "superglue-fast", "NN-superpoint", "NN-ratio", "NN-mutual", "adalam"],
Literal["NN", "superglue", "superglue-fast", "NN-superpoint", "NN-ratio", "NN-mutual", "adalam"],
],
]:
"""Find a valid combination of sfm tool, feature type, and matcher type.
@@ -452,7 +452,7 @@ def generate_circle_mask(height: int, width: int, percent_radius) -> Optional[np
mask = np.zeros((height, width), dtype=np.uint8)
center = (width // 2, height // 2)
radius = int(percent_radius * np.sqrt(width**2 + height**2) / 2.0)
cv2.circle(mask, center, radius, 1, -1) # type: ignore
cv2.circle(mask, center, radius, 1, -1)
return mask
@@ -523,7 +523,7 @@ def save_mask(
The path to the mask file or None if no mask is needed.
"""
image_path = next(image_dir.glob("frame_*"))
image = cv2.imread(str(image_path)) # type: ignore
image = cv2.imread(str(image_path))
height, width = image.shape[:2]
mask = generate_mask(height, width, crop_factor, percent_radius)
if mask is None:
@@ -531,17 +531,17 @@ def save_mask(
mask *= 255
mask_path = image_dir.parent / "masks"
mask_path.mkdir(exist_ok=True)
cv2.imwrite(str(mask_path / "mask.png"), mask) # type: ignore
cv2.imwrite(str(mask_path / "mask.png"), mask)
downscale_factors = [2**i for i in range(num_downscales + 1)[1:]]
for downscale in downscale_factors:
mask_path_i = image_dir.parent / f"masks_{downscale}"
mask_path_i.mkdir(exist_ok=True)
mask_path_i = mask_path_i / "mask.png"
mask_i = cv2.resize( # type: ignore
mask_i = cv2.resize(
mask,
(width // downscale, height // downscale),
interpolation=cv2.INTER_NEAREST, # type: ignore
interpolation=cv2.INTER_NEAREST,
)
cv2.imwrite(str(mask_path_i), mask_i) # type: ignore
cv2.imwrite(str(mask_path_i), mask_i)
CONSOLE.log(":tada: Generated and saved masks.")
return mask_path / "mask.png"
@@ -17,7 +17,7 @@
import csv
import json
from pathlib import Path
from typing import List
from typing import Dict, List
import numpy as np
from PIL import Image
@@ -27,7 +27,7 @@ from nerfstudio.utils.rich_utils import CONSOLE
def realitycapture_to_json(
image_filename_map: List[Path],
image_filename_map: Dict[str, Path],
csv_filename: Path,
output_dir: Path,
verbose: bool = False,
@@ -100,15 +100,17 @@ class ProcessNuScenesMasks:
mask[-100:] = 0
for box in boxes:
# Dont mask out static objects (static in all frames)
# Dont' mask out static objects (static in all frames)
instance_token = nusc.get("sample_annotation", box.token)["instance_token"]
if not instances_is_dynamic[instance_token]:
continue
# project box to image plane and rasterize each face
# Project box to image plane and rasterize each face
corners_3d = box.corners()
corners = view_points(corners_3d, intrinsics, normalize=True)[:2, :]
corners = np.round(corners).astype(int).T
# Type ignores needed because fillPoly expects cv2.Mat
cv2.fillPoly(mask, [corners[[0, 1, 2, 3]]], 0) # front # type: ignore
cv2.fillPoly(mask, [corners[[4, 5, 6, 7]]], 0) # back # type: ignore
cv2.fillPoly(mask, [corners[[0, 1, 5, 4]]], 0) # top # type: ignore
@@ -117,27 +119,30 @@ class ProcessNuScenesMasks:
cv2.fillPoly(mask, [corners[[1, 2, 6, 5]]], 0) # right # type: ignore
maskname = os.path.split(camera_data["filename"])[1].replace("jpg", "png")
cv2.imwrite(str(self.output_dir / "masks" / camera / maskname), mask * 255) # type: ignore
cv2.imwrite(
str(self.output_dir / "masks" / camera / maskname),
mask * 255, # type: ignore
)
if self.verbose:
img = cv2.imread(str(self.data_dir / camera_data["filename"])) # type: ignore
img = cv2.imread(str(self.data_dir / camera_data["filename"]))
mask = ~mask.astype(bool)
img[mask, :] -= np.minimum(img[mask, :], 100)
img[mask, :] = img[mask, :] - np.minimum(img[mask, :], 100)
viz.append(img)
if self.verbose:
if len(viz) == 6:
viz = np.vstack((np.hstack(viz[:3]), np.hstack(viz[3:])))
viz = cv2.resize(viz, (int(1600 * 3 / 3), int(900 * 2 / 3))) # type: ignore
viz = cv2.resize(viz, (int(1600 * 3 / 3), int(900 * 2 / 3)))
elif len(viz) == 3:
viz = np.hstack(viz[:3])
viz = cv2.resize(viz, (int(1600 * 3 / 3), int(900 / 3))) # type: ignore
viz = cv2.resize(viz, (int(1600 * 3 / 3), int(900 / 3)))
elif len(viz) == 1:
viz = viz[0]
else:
raise ValueError("Only support 1 or 3 or 6 cameras for viz")
cv2.imshow("", viz) # type: ignore
cv2.waitKey(1) # type: ignore
cv2.imshow("", viz)
cv2.waitKey(1)
def entrypoint():
+1 -1
View File
@@ -23,7 +23,7 @@ from rich.style import Style
from nerfstudio.utils.rich_utils import CONSOLE
LOCAL_TESTS = ["Run license checks", "Run Ruff", "Run Black", "Test with pytest"]
LOCAL_TESTS = ["Run license checks", "Run Ruff", "Run Black", "Run Pyright", "Test with pytest"]
def run_command(command: str, continue_on_fail: bool = False) -> bool:
+2
View File
@@ -51,6 +51,7 @@ from nerfstudio.cameras.camera_paths import (
get_spiral_path,
)
from nerfstudio.cameras.cameras import Cameras, CameraType
from nerfstudio.data.datamanagers.base_datamanager import VanillaDataManager
from nerfstudio.data.scene_box import SceneBox
from nerfstudio.model_components import renderers
from nerfstudio.pipelines.base_pipeline import Pipeline
@@ -416,6 +417,7 @@ class SpiralRender(BaseRender):
install_checks.check_ffmpeg_installed()
assert isinstance(pipeline.datamanager, VanillaDataManager)
steps = int(self.frame_rate * self.seconds)
camera_start = pipeline.datamanager.eval_dataloader.get_camera(image_idx=0).flatten()
camera_path = get_spiral_path(camera_start, steps=steps, radius=self.radius)
+7 -1
View File
@@ -79,6 +79,12 @@ dev = [
"typeguard==2.13.3",
"ruff==0.0.267",
"sshconf==0.2.5",
"pycolmap==0.3.0",
"hloc@git+https://github.com/cvg/Hierarchical-Localization.git",
"diffusers==0.16.1",
"opencv-stubs==0.0.7",
"transformers==4.29.2",
"pyright==1.1.308",
]
# Documentation related packages
@@ -139,7 +145,7 @@ exclude = ["**/node_modules",
]
defineConstant = { DEBUG = true }
reportMissingImports = true
reportMissingImports = "warning"
reportMissingTypeStubs = false
reportPrivateImportUsage = false
@@ -15,9 +15,10 @@ def test_dnerf_distortion():
num_rays = 1024
num_samples = 256
positions = torch.rand((num_rays, num_samples, 3), dtype=torch.float32, device=device)
assert distortion.forward(positions, None) is None
times = torch.rand_like(positions[..., :1])
assert distortion.forward(positions, times).shape == positions.shape
out = distortion.forward(positions, times)
assert out is not None
assert out.shape == positions.shape
if __name__ == "__main__":
@@ -34,6 +34,7 @@ def test_temporal_grid():
weight = torch.randn_like(out)
(out * weight).sum().backward()
assert torch.all(out == 0.5)
assert model.embeddings.grad is not None
assert model.embeddings.grad.sum() - weight.sum() < 0.01
assert torch.all(model.embeddings.grad[2:, :] == 0)
assert torch.all(model.embeddings.grad[:, 1:] == 0)
+2 -2
View File
@@ -75,7 +75,7 @@ def test_load_state_dict():
state_dict["_model.param"].mul_(2)
pipeline.load_pipeline(state_dict, 0)
assert was_called
assert pipeline.model.param[0].item() == 2
assert getattr(pipeline.model, "param")[0].item() == 2
# preparation for another test
state_dict["_model.param"].mul_(2)
@@ -91,4 +91,4 @@ def test_load_state_dict():
# load DDP checkpoint
pipeline.load_pipeline(ddp_state_dict, 0)
assert was_called
assert pipeline.model.param[0].item() == 4
assert getattr(pipeline.model, "param")[0].item() == 4
+3
View File
@@ -12,6 +12,7 @@ from nerfstudio.configs.method_configs import method_configs
from nerfstudio.data.dataparsers.blender_dataparser import BlenderDataParserConfig
from nerfstudio.data.dataparsers.minimal_dataparser import MinimalDataParserConfig
from nerfstudio.engine.trainer import TrainerConfig
from nerfstudio.models.vanilla_nerf import VanillaModelConfig
from nerfstudio.scripts.train import train_loop
BLACKLIST = [
@@ -46,8 +47,10 @@ def set_reduced_config(config: TrainerConfig):
# reduce model factors
if hasattr(config.pipeline.model, "num_coarse_samples"):
assert isinstance(config.pipeline.model, VanillaModelConfig)
config.pipeline.model.num_coarse_samples = 4
if hasattr(config.pipeline.model, "num_importance_samples"):
assert isinstance(config.pipeline.model, VanillaModelConfig)
config.pipeline.model.num_importance_samples = 4
# remove viewer
config.viewer.quit_on_train_completion = True
+15 -11
View File
@@ -1,8 +1,8 @@
"""
Test tensor dataclass
"""
from dataclasses import dataclass
from typing import Dict
from dataclasses import dataclass, field
from typing import Generic, Optional, TypeVar
import pytest
import torch
@@ -17,14 +17,17 @@ class DummyNestedClass(TensorDataclass):
x: torch.Tensor
MaybeTensorDataclass = TypeVar("MaybeTensorDataclass")
@dataclass
class DummyTensorDataclass(TensorDataclass):
class DummyTensorDataclass(TensorDataclass, Generic[MaybeTensorDataclass]):
"""Dummy dataclass"""
a: torch.Tensor
b: torch.Tensor
c: DummyNestedClass = None
d: Dict = None
c: MaybeTensorDataclass
d: dict = field(default_factory=dict)
def test_init():
@@ -34,7 +37,7 @@ def test_init():
class Dummy(TensorDataclass):
"""Dummy dataclass"""
dummy_vals: torch.Tensor = None
dummy_vals: Optional[torch.Tensor] = None
Dummy(dummy_vals=torch.ones(1))
with pytest.raises(ValueError):
@@ -46,19 +49,19 @@ def test_broadcasting():
a = torch.ones((4, 6, 3))
b = torch.ones((6, 2))
tensor_dataclass = DummyTensorDataclass(a=a, b=b)
tensor_dataclass = DummyTensorDataclass(a=a, b=b, c=None)
assert tensor_dataclass.b.shape == (4, 6, 2)
a = torch.ones((4, 6, 3))
b = torch.ones(2)
tensor_dataclass = DummyTensorDataclass(a=a, b=b)
tensor_dataclass = DummyTensorDataclass(a=a, b=b, c=None)
assert tensor_dataclass.b.shape == (4, 6, 2)
# Invalid broadcasting
a = torch.ones((4, 6, 3))
b = torch.ones((3, 2))
with pytest.raises(RuntimeError):
tensor_dataclass = DummyTensorDataclass(a=a, b=b)
tensor_dataclass = DummyTensorDataclass(a=a, b=b, c=None)
def test_tensor_ops():
@@ -67,7 +70,7 @@ def test_tensor_ops():
a = torch.ones((4, 6, 3))
b = torch.ones((6, 2))
d = {"t1": torch.ones((4, 6, 3)), "t2": {"t3": torch.ones((6, 4))}}
tensor_dataclass = DummyTensorDataclass(a=a, b=b, d=d)
tensor_dataclass = DummyTensorDataclass(a=a, b=b, c=None, d=d)
assert tensor_dataclass.shape == (4, 6)
assert tensor_dataclass.a.shape == (4, 6, 3)
@@ -114,6 +117,7 @@ def test_tensor_ops():
tensor_dataclass = DummyTensorDataclass(
a=torch.ones((2, 3, 4, 5)),
b=torch.ones((4, 5)),
c=None,
d={"t1": torch.ones((2, 3, 4, 5)), "t2": {"t3": torch.ones((4, 5))}},
)
assert tensor_dataclass[0, ...].shape == (3, 4)
@@ -168,7 +172,7 @@ def test_nested_class():
def test_iter():
"""Test iterating over tensor dataclass"""
tensor_dataclass = DummyTensorDataclass(a=torch.ones((3, 4, 5)), b=torch.ones((3, 4, 5)))
tensor_dataclass = DummyTensorDataclass(a=torch.ones((3, 4, 5)), b=torch.ones((3, 4, 5)), c=None)
for batch in tensor_dataclass:
assert batch.shape == (4,)
assert batch.a.shape == (4, 5)