xrfeitoria.utils.viewer

Utils for loading images and annotations.

class xrfeitoria.utils.viewer.Viewer(sequence_dir: str | Path)

Utils for loading images and annotations.

Examples

>>> viewer = Viewer('/path/to/sequence')
>>> camera_name = 'cam'
>>> frame = 0
>>> img = viewer.get_img(camera_name=camera_name, frame=frame)
>>> mask = viewer.get_mask(camera_name=camera_name, frame=frame)
>>> depth = viewer.get_depth(camera_name=camera_name, frame=frame)
>>> flow = viewer.get_flow(camera_name=camera_name, frame=frame)
>>> normal = viewer.get_normal(camera_name=camera_name, frame=frame)
>>> diffuse = viewer.get_diffuse(camera_name=camera_name, frame=frame)

__init__(sequence_dir: str | Path) → None

Initialize with the sequence directory.

Parameters:

sequence_dir (PathLike) – path to the sequence directory

get_depth(camera_name: str, frame: int, inverse: bool = False, depth_rescale: float = 1.0) → ndarray

Get depth of the given frame (‘depth/{frame:04d}.*’)

Parameters:

• camera_name (str) – the camera name

• frame (int) – the frame number

• inverse (bool, optional) – whether to inverse the depth. If True, white (255) represents the farthest, and black (0) represents the nearest. if False, white (255) represents the nearest, and black (0) represents the farthest. Defaults to False.

• depth_rescale (float, optional) – scaling the depth to map it into (0, 255). depth = depth / depth_rescale. Depth values greater than depth_rescale will be clipped. Defaults to 1.0.

Returns:

np.ndarray – depth of shape (H, W, 3)

get_diffuse(camera_name: str, frame: int) → ndarray

Get diffuse image of the given frame (‘diffuse/{frame:04d}.*’)

Parameters:

• camera_name (str) – the camera name

• frame (int) – the frame number

Returns:

np.ndarray – image of shape (H, W, 3)

get_flow(camera_name: str, frame: int) → ndarray

Get optical flow of the given frame (‘flow/{frame:04d}.*’)

Parameters:

• camera_name (str) – the camera name

• frame (int) – the frame number

Returns:

np.ndarray – optical flow of shape (H, W, 3)

get_img(camera_name: str, frame: int) → ndarray

Get rgb image of the given frame (‘img/{frame:04d}.*’)

Parameters:

• camera_name (str) – the camera name

• frame (int) – the frame number

Returns:

np.ndarray – image of shape (H, W, 3)

get_mask(camera_name: str, frame: int) → ndarray

Get mask of the given frame (‘mask/{frame:04d}.*’)

Parameters:

• camera_name (str) – the camera name

• frame (int) – the frame number

Returns:

np.ndarray – image of shape (H, W, 3)

get_normal(camera_name: str, frame: int) → ndarray

Get normal map of the given frame (‘normal/{frame:04d}.*’)

Parameters:

• camera_name (str) – the camera name

• frame (int) – the frame number

Returns:

np.ndarray – normal map of shape (H, W, 3)

DEPTH = 'depth'

DIFFUSE = 'diffuse'

FLOW = 'flow'

IMG = 'img'

MASK = 'mask'

NORMAL = 'normal'

property camera_names: List[str]

property frame_num: int