nextcv.sensors.camera ⚓︎

Sensor representation and manipulation utilities.

CLASS	DESCRIPTION
`Camera`	Represents a camera with its intrinsics, pose, and image resolution.
`PinholeCamera`	Represents a pinhole camera with its intrinsics and pose.

nextcv.sensors.camera.Camera ⚓︎

Bases: BaseModel

Represents a camera with its intrinsics, pose, and image resolution.

PARAMETER	DESCRIPTION
`width`	Image width in pixels TYPE: `int` DEFAULT: `None`
`height`	Image height in pixels TYPE: `int` DEFAULT: `None`
`fx`	Focal length along x-axis in pixels TYPE: `float` DEFAULT: `None`
`fy`	Focal length along y-axis in pixels TYPE: `float` DEFAULT: `None`
`cx`	Principal point x-coordinate in pixels TYPE: `float` DEFAULT: `None`
`cy`	Principal point y-coordinate in pixels TYPE: `float` DEFAULT: `None`
`roll`	Camera roll angle in degrees (rotation around z-axis) TYPE: `float` DEFAULT: `None`
`pitch`	Camera pitch angle in degrees (rotation around x-axis) TYPE: `float` DEFAULT: `None`
`yaw`	Camera yaw angle in degrees (rotation around y-axis) TYPE: `float` DEFAULT: `None`

METHOD	DESCRIPTION
`compute_homography_to`	Compute homography matrix that maps points from this cam to target cam.
`from_dict`	Create a Camera instance from intrinsics and pose dictionaries.
`maps_from`	Create remapping maps from source camera to this camera.
`must_be_even`	Ensure width and height are positive integers.

ATTRIBUTE	DESCRIPTION
`size`	Get the camera size as an opencv-compatible tuple of width and height. TYPE: `Tuple[int, int]`

size `property` ⚓︎

size: Tuple[int, int]

Get the camera size as an opencv-compatible tuple of width and height.

compute_homography_to ⚓︎

compute_homography_to(target: Camera, neg_focal_length: bool = True) -> ndarray

Compute homography matrix that maps points from this cam to target cam.

from_dict `classmethod` ⚓︎

from_dict(data: Dict[str, Any]) -> T

Create a Camera instance from intrinsics and pose dictionaries.

PARAMETER	DESCRIPTION
`data`	Dictionary with keys "fx", "fy", "cx", "cy", "width", "height", "roll", "pitch", "yaw" TYPE: `Dict[str, Any]`

RETURNS	DESCRIPTION
`T`	Camera instance of the correct subclass type

maps_from ⚓︎

maps_from(src: Camera, neg_focal_length: bool = True) -> Tuple[ndarray, ndarray]

Create remapping maps from source camera to this camera.

must_be_even ⚓︎

must_be_even(v: int) -> int

Ensure width and height are positive integers.

nextcv.sensors.camera.PinholeCamera ⚓︎

Bases: Camera

Represents a pinhole camera with its intrinsics and pose.

PARAMETER	DESCRIPTION
`width`	Image width in pixels TYPE: `int` DEFAULT: `None`
`height`	Image height in pixels TYPE: `int` DEFAULT: `None`
`fx`	Focal length along x-axis in pixels TYPE: `float` DEFAULT: `None`
`fy`	Focal length along y-axis in pixels TYPE: `float` DEFAULT: `None`
`cx`	Principal point x-coordinate in pixels TYPE: `float` DEFAULT: `None`
`cy`	Principal point y-coordinate in pixels TYPE: `float` DEFAULT: `None`
`roll`	Camera roll angle in degrees (rotation around z-axis) TYPE: `float` DEFAULT: `None`
`pitch`	Camera pitch angle in degrees (rotation around x-axis) TYPE: `float` DEFAULT: `None`
`yaw`	Camera yaw angle in degrees (rotation around y-axis) TYPE: `float` DEFAULT: `None`

METHOD	DESCRIPTION
`compute_homography_to`	Compute homography matrix that maps points from this camera to target camera.
`crop`	Return a new PinholeCamera cropped by margins (left, top, right, bottom).
`hconcat`	Create a virtual camera by concatenating left and right cams horizontally.
`maps_from`	Create remapping maps from source camera to this camera.
`vconcat`	Create a virtual camera by concatenating top and bottom cams vertically.

ATTRIBUTE	DESCRIPTION
`K`	Get the camera intrinsics matrix. TYPE: `ndarray`
`R`	Get the camera rotation matrix from Euler angles with pinhole convention. TYPE: `ndarray`

K `property` ⚓︎

K: ndarray

Get the camera intrinsics matrix.

R `property` ⚓︎

R: ndarray

Get the camera rotation matrix from Euler angles with pinhole convention.

The pinhole convention is: - x is right - y is down - z is forward

compute_homography_to ⚓︎

compute_homography_to(target: PinholeCamera, neg_focal_length: bool = True) -> ndarray

Compute homography matrix that maps points from this camera to target camera.

PARAMETER	DESCRIPTION
`target`	Target camera to compute homography to TYPE: `PinholeCamera`
`neg_focal_length`	Whether to negate the focal length TYPE: `bool` DEFAULT: `True`

RETURNS	DESCRIPTION
`ndarray`	Homography matrix that transforms points from this camera to target camera

crop ⚓︎

crop(left: float, top: float, right: float, bottom: float, force_even: bool = True) -> PinholeCamera

Return a new PinholeCamera cropped by margins (left, top, right, bottom).

hconcat `classmethod` ⚓︎

hconcat(left: PinholeCamera, right: PinholeCamera) -> PinholeCamera

Create a virtual camera by concatenating left and right cams horizontally.

maps_from ⚓︎

maps_from(src: PinholeCamera, neg_focal_length: bool = True) -> Tuple[ndarray, ndarray]

Create remapping maps from source camera to this camera.

vconcat `classmethod` ⚓︎

vconcat(top: PinholeCamera, bottom: PinholeCamera) -> PinholeCamera

Create a virtual camera by concatenating top and bottom cams vertically.

nextcv.sensors.camera ⚓︎

nextcv.sensors.camera.Camera ⚓︎

size property ⚓︎

compute_homography_to ⚓︎

from_dict classmethod ⚓︎

maps_from ⚓︎

must_be_even ⚓︎

nextcv.sensors.camera.PinholeCamera ⚓︎

K property ⚓︎

R property ⚓︎

compute_homography_to ⚓︎

crop ⚓︎

hconcat classmethod ⚓︎

maps_from ⚓︎

vconcat classmethod ⚓︎

size `property` ⚓︎

from_dict `classmethod` ⚓︎

K `property` ⚓︎

R `property` ⚓︎

hconcat `classmethod` ⚓︎

vconcat `classmethod` ⚓︎