nerfbaselines.datasets

class nerfbaselines.datasets.Dataset

Bases: _IncompleteDataset

cameras: Cameras

image_paths: List[str]

image_paths_root: str

images: ndarray | List[ndarray]

images_points3D_indices: List[ndarray] | None

metadata: Dict

points3D_rgb: ndarray | None

points3D_xyz: ndarray | None

sampling_mask_paths: List[str] | None

sampling_mask_paths_root: str | None

sampling_masks: ndarray | List[ndarray] | None

exception nerfbaselines.datasets.DatasetNotFoundError

Bases: Exception

exception nerfbaselines.datasets.MultiDatasetError(errors, message)

Bases: DatasetNotFoundError

write_to_logger(color=True, terminal_width=None)

class nerfbaselines.datasets.UnloadedDataset

Bases: _IncompleteDataset

cameras: Cameras

image_paths: List[str]

image_paths_root: str

images: NotRequired[ndarray | List[ndarray] | None]

images_points3D_indices: List[ndarray] | None

metadata: Dict

points3D_rgb: ndarray | None

points3D_xyz: ndarray | None

sampling_mask_paths: List[str] | None

sampling_mask_paths_root: str | None

sampling_masks: ndarray | List[ndarray] | None

nerfbaselines.datasets.dataset_index_select(dataset: TDataset, i: slice | int | list | ndarray) → TDataset

nerfbaselines.datasets.dataset_load_features(dataset: UnloadedDataset, features=None, supported_camera_models=None) → Dataset

nerfbaselines.datasets.download_dataset(path: str, output: str | Path)

nerfbaselines.datasets.load_dataset(path: Path | str, split: str, features: FrozenSet[Literal['color', 'points3D_xyz', 'points3D_rgb']] | None = None, supported_camera_models: FrozenSet[Literal['pinhole', 'opencv', 'opencv_fisheye', 'full_opencv']] | None = None, load_features: Literal[True] = True, **kwargs) → Dataset

nerfbaselines.datasets.load_dataset(path: Path | str, split: str, features: FrozenSet[Literal['color', 'points3D_xyz', 'points3D_rgb']] | None = None, supported_camera_models: FrozenSet[Literal['pinhole', 'opencv', 'opencv_fisheye', 'full_opencv']] | None = None, load_features: Literal[False] = True, **kwargs) → UnloadedDataset

nerfbaselines.datasets.new_dataset(*, cameras: Cameras, image_paths: Sequence[str], image_paths_root: str | None = None, images: ndarray | List[ndarray] | None = None, sampling_mask_paths: Sequence[str] | None = None, sampling_mask_paths_root: str | None = None, sampling_masks: ndarray | List[ndarray] | None = None, points3D_xyz: ndarray | None = None, points3D_rgb: ndarray | None = None, images_points3D_indices: Sequence[ndarray] | None = None, metadata: Dict) → UnloadedDataset | Dataset

nerfbaselines.datasets.blender

nerfbaselines.datasets.blender.camera_model_to_int(camera_model: Literal['pinhole', 'opencv', 'opencv_fisheye', 'full_opencv']) → int

nerfbaselines.datasets.blender.download_blender_dataset(path: str, output: Path)

nerfbaselines.datasets.blender.get_default_viewer_transform(poses, dataset_type: str | None) → Tuple[ndarray, ndarray]

nerfbaselines.datasets.blender.load_blender_dataset(path: Path | str, split: str, **kwargs)

nerfbaselines.datasets.blender.new_cameras(*, poses: ndarray, intrinsics: ndarray, camera_types: ndarray, distortion_parameters: ndarray, image_sizes: ndarray, nears_fars: ndarray | None = None, metadata: ndarray | None = None) → Cameras

nerfbaselines.datasets.bundler

nerfbaselines.datasets.bundler.camera_model_to_int(camera_model: Literal['pinhole', 'opencv', 'opencv_fisheye', 'full_opencv']) → int

nerfbaselines.datasets.bundler.get_default_viewer_transform(poses, dataset_type: str | None) → Tuple[ndarray, ndarray]

nerfbaselines.datasets.bundler.get_split_and_train_indices(image_names, path, split)

nerfbaselines.datasets.bundler.load_bundler_dataset(path: str, split: str | None = None, images_path: str | None = None, features: FrozenSet[Literal['color', 'points3D_xyz', 'points3D_rgb']] | None = None, sampling_masks_path: str | None = None)

nerfbaselines.datasets.bundler.load_bundler_file(path: str, image_list: List[str])

nerfbaselines.datasets.bundler.new_cameras(*, poses: ndarray, intrinsics: ndarray, camera_types: ndarray, distortion_parameters: ndarray, image_sizes: ndarray, nears_fars: ndarray | None = None, metadata: ndarray | None = None) → Cameras

nerfbaselines.datasets.colmap

class nerfbaselines.datasets.colmap.Camera(id, model, width, height, params)

Bases: tuple

class nerfbaselines.datasets.colmap.Image(id, qvec, tvec, camera_id, name, xys, point3D_ids)

Bases: BaseImage

qvec2rotmat()

class nerfbaselines.datasets.colmap.Point3D(id, xyz, rgb, error, image_ids, point2D_idxs)

Bases: tuple

nerfbaselines.datasets.colmap.camera_model_to_int(camera_model: Literal['pinhole', 'opencv', 'opencv_fisheye', 'full_opencv']) → int

nerfbaselines.datasets.colmap.get_default_viewer_transform(poses, dataset_type: str | None) → Tuple[ndarray, ndarray]

nerfbaselines.datasets.colmap.load_colmap_dataset(path: Path | str, split: str | None = None, *, test_indices: Indices | None = None, features: FrozenSet[Literal['color', 'points3D_xyz', 'points3D_rgb']] | None = None, images_path: Path | str | None = None, colmap_path: Path | str | None = None, sampling_masks_path: Path | str | None = None)

nerfbaselines.datasets.colmap.new_cameras(*, poses: ndarray, intrinsics: ndarray, camera_types: ndarray, distortion_parameters: ndarray, image_sizes: ndarray, nears_fars: ndarray | None = None, metadata: ndarray | None = None) → Cameras

nerfbaselines.datasets.colmap.padded_stack(tensors: ndarray | Tuple[ndarray, ...] | List[ndarray]) → ndarray

nerfbaselines.datasets.colmap.qvec2rotmat(qvec)

nerfbaselines.datasets.colmap.read_cameras_binary(path_to_model_file)

see: src/base/reconstruction.cc

void Reconstruction::WriteCamerasBinary(const std::string& path) void Reconstruction::ReadCamerasBinary(const std::string& path)

nerfbaselines.datasets.colmap.read_cameras_text(path)

see: src/base/reconstruction.cc

void Reconstruction::WriteCamerasText(const std::string& path) void Reconstruction::ReadCamerasText(const std::string& path)

nerfbaselines.datasets.colmap.read_images_binary(path_to_model_file)

see: src/base/reconstruction.cc

void Reconstruction::ReadImagesBinary(const std::string& path) void Reconstruction::WriteImagesBinary(const std::string& path)

nerfbaselines.datasets.colmap.read_images_text(path)

see: src/base/reconstruction.cc

void Reconstruction::ReadImagesText(const std::string& path) void Reconstruction::WriteImagesText(const std::string& path)

nerfbaselines.datasets.colmap.read_points3D_binary(path_to_model_file)

see: src/base/reconstruction.cc

void Reconstruction::ReadPoints3DBinary(const std::string& path) void Reconstruction::WritePoints3DBinary(const std::string& path)

nerfbaselines.datasets.colmap.read_points3D_text(path)

see: src/base/reconstruction.cc

void Reconstruction::ReadPoints3DText(const std::string& path) void Reconstruction::WritePoints3DText(const std::string& path)

nerfbaselines.datasets.llff

nerfbaselines.datasets.llff.camera_model_to_int(camera_model: Literal['pinhole', 'opencv', 'opencv_fisheye', 'full_opencv']) → int

nerfbaselines.datasets.llff.download_llff_dataset(path: str, output: Path | str)

nerfbaselines.datasets.llff.load_llff_dataset(path: Path | str, split: str, downscale_factor: int = 4, **_)

nerfbaselines.datasets.llff.new_cameras(*, poses: ndarray, intrinsics: ndarray, camera_types: ndarray, distortion_parameters: ndarray, image_sizes: ndarray, nears_fars: ndarray | None = None, metadata: ndarray | None = None) → Cameras

nerfbaselines.datasets.mipnerf360

nerfbaselines.datasets.mipnerf360.download_mipnerf360_dataset(path: str, output: Path | str)

nerfbaselines.datasets.mipnerf360.get_default_viewer_transform(poses, dataset_type: str | None) → Tuple[ndarray, ndarray]

nerfbaselines.datasets.mipnerf360.get_scene_scale(cameras: Cameras, dataset_type: Literal['object-centric', 'forward-facing'] | None)

nerfbaselines.datasets.mipnerf360.load_colmap_dataset(path: Path | str, split: str | None = None, *, test_indices: Indices | None = None, features: FrozenSet[Literal['color', 'points3D_xyz', 'points3D_rgb']] | None = None, images_path: Path | str | None = None, colmap_path: Path | str | None = None, sampling_masks_path: Path | str | None = None)

nerfbaselines.datasets.mipnerf360.load_mipnerf360_dataset(path: Path | str, split: str, resize_full_image: bool = False, downscale_factor=None, **kwargs)

nerfbaselines.datasets.mipnerf360.single(xs)

nerfbaselines.datasets.nerfstudio

nerfbaselines.datasets.nerfstudio.camera_model_to_int(camera_model: Literal['pinhole', 'opencv', 'opencv_fisheye', 'full_opencv']) → int

nerfbaselines.datasets.nerfstudio.download_capture_name(output: Path, file_id_or_zip_url)

Download specific captures a given dataset and capture name.

nerfbaselines.datasets.nerfstudio.download_nerfstudio_dataset(path: str, output: Path | str)

Download data in the Nerfstudio format. If you are interested in the Nerfstudio Dataset subset from the SIGGRAPH 2023 paper, you can obtain that by using –capture-name nerfstudio-dataset or by visiting Google Drive directly at: https://drive.google.com/drive/folders/19TV6kdVGcmg3cGZ1bNIUnBBMD-iQjRbG?usp=drive_link.

nerfbaselines.datasets.nerfstudio.get_default_viewer_transform(poses, dataset_type: str | None) → Tuple[ndarray, ndarray]

nerfbaselines.datasets.nerfstudio.get_scene_scale(cameras: Cameras, dataset_type: Literal['object-centric', 'forward-facing'] | None)

nerfbaselines.datasets.nerfstudio.get_train_eval_split_all(image_filenames: List) → Tuple[ndarray, ndarray]

Get the train/eval split where all indices are used for both train and eval.

Parameters:

image_filenames – list of image filenames

nerfbaselines.datasets.nerfstudio.get_train_eval_split_filename(image_filenames: List) → Tuple[ndarray, ndarray]

Get the train/eval split based on the filename of the images.

Parameters:

image_filenames – list of image filenames

nerfbaselines.datasets.nerfstudio.get_train_eval_split_fraction(image_filenames: List, train_split_fraction: float) → Tuple[ndarray, ndarray]

Get the train/eval split fraction based on the number of images and the train split fraction.

Parameters:

• image_filenames – list of image filenames

• train_split_fraction – fraction of images to use for training

nerfbaselines.datasets.nerfstudio.get_train_eval_split_interval(image_filenames: List, eval_interval: float) → Tuple[ndarray, ndarray]

Get the train/eval split based on the interval of the images.

Parameters:

• image_filenames – list of image filenames

• eval_interval – interval of images to use for eval

nerfbaselines.datasets.nerfstudio.grab_file_id(zip_url: str) → str

Get the file id from the google drive zip url.

nerfbaselines.datasets.nerfstudio.load_from_json(filename: Path)

Load a dictionary from a JSON filename.

Parameters:

filename – The filename to load from.

nerfbaselines.datasets.nerfstudio.load_nerfstudio_dataset(path: Path | str, split: str, downscale_factor: int | None = None, features: FrozenSet[Literal['color', 'points3D_xyz', 'points3D_rgb']] | None = None, **kwargs)

nerfbaselines.datasets.nerfstudio.new_cameras(*, poses: ndarray, intrinsics: ndarray, camera_types: ndarray, distortion_parameters: ndarray, image_sizes: ndarray, nears_fars: ndarray | None = None, metadata: ndarray | None = None) → Cameras

nerfbaselines.datasets.nerfstudio.read_images_binary(path_to_model_file)

see: src/base/reconstruction.cc

void Reconstruction::ReadImagesBinary(const std::string& path) void Reconstruction::WriteImagesBinary(const std::string& path)

nerfbaselines.datasets.nerfstudio.read_images_text(path)

see: src/base/reconstruction.cc

void Reconstruction::ReadImagesText(const std::string& path) void Reconstruction::WriteImagesText(const std::string& path)

nerfbaselines.datasets.nerfstudio.read_points3D_binary(path_to_model_file)

see: src/base/reconstruction.cc

void Reconstruction::ReadPoints3DBinary(const std::string& path) void Reconstruction::WritePoints3DBinary(const std::string& path)

nerfbaselines.datasets.nerfstudio.read_points3D_text(path)

see: src/base/reconstruction.cc

void Reconstruction::ReadPoints3DText(const std::string& path) void Reconstruction::WritePoints3DText(const std::string& path)

nerfbaselines.datasets.nerfstudio.wget(url: str, output: str | Path)

nerfbaselines.datasets.phototourism

class nerfbaselines.datasets.phototourism.EvaluationProtocol(*args, **kwargs)

Bases: Protocol

accumulate_metrics(metrics: Iterable[Dict[str, float | int]]) → Dict[str, float | int]

evaluate(predictions: Iterable[RenderOutput], dataset: Dataset) → Iterable[Dict[str, float | int]]

get_name() → str

render(method: Method, dataset: Dataset) → Iterable[RenderOutput]

class nerfbaselines.datasets.phototourism.NerfWEvaluationProtocol

Bases: EvaluationProtocol

accumulate_metrics(metrics: Iterable[Dict[str, float | int]]) → Dict[str, float | int]

evaluate(predictions: Iterable[RenderOutput], dataset: Dataset) → Iterable[Dict[str, float | int]]

get_name()

render(method: Method, dataset: Dataset) → Iterable[RenderOutput]

nerfbaselines.datasets.phototourism.download_phototourism_dataset(path: str, output: Path | str)

nerfbaselines.datasets.phototourism.get_default_viewer_transform(poses, dataset_type: str | None) → Tuple[ndarray, ndarray]

nerfbaselines.datasets.phototourism.get_scene_scale(cameras: Cameras, dataset_type: Literal['object-centric', 'forward-facing'] | None)

nerfbaselines.datasets.phototourism.horizontal_half_dataset(dataset: Dataset, left: bool = True) → Dataset

nerfbaselines.datasets.phototourism.image_to_srgb(tensor, dtype, color_space: str | None = None, allow_alpha: bool = False, background_color: ndarray | None = None)

nerfbaselines.datasets.phototourism.load_colmap_dataset(path: Path | str, split: str | None = None, *, test_indices: Indices | None = None, features: FrozenSet[Literal['color', 'points3D_xyz', 'points3D_rgb']] | None = None, images_path: Path | str | None = None, colmap_path: Path | str | None = None, sampling_masks_path: Path | str | None = None)

nerfbaselines.datasets.phototourism.load_phototourism_dataset(path: Path | str, split: str, use_nerfw_split=None, **kwargs)

nerfbaselines.datasets.phototourism.single(xs)

nerfbaselines.datasets.tanksandtemples

nerfbaselines.datasets.tanksandtemples.assert_not_none(value: T | None) → T

nerfbaselines.datasets.tanksandtemples.download_tanksandtemples_dataset(path: str, output: Path | str) → None

nerfbaselines.datasets.tanksandtemples.get_default_viewer_transform(poses, dataset_type: str | None) → Tuple[ndarray, ndarray]

nerfbaselines.datasets.tanksandtemples.get_scene_scale(cameras: Cameras, dataset_type: Literal['object-centric', 'forward-facing'] | None)

nerfbaselines.datasets.tanksandtemples.load_colmap_dataset(path: Path | str, split: str | None = None, *, test_indices: Indices | None = None, features: FrozenSet[Literal['color', 'points3D_xyz', 'points3D_rgb']] | None = None, images_path: Path | str | None = None, colmap_path: Path | str | None = None, sampling_masks_path: Path | str | None = None)

nerfbaselines.datasets.tanksandtemples.load_tanksandtemples_dataset(path: Path | str, split: str, downscale_factor: int = 2, **kwargs) → UnloadedDataset