pypose.reprojerr

class pypose.reprojerr(points, pixels, intrinsics, extrinsics)

Calculates batched per-pixel reprojection error (pixel distance) for points in the world coordinate, given camera intrinsic and extrinsic parameters.

Parameters:

• points (torch.Tensor) – The object points in world coordinate. The shape has to be (…, N, 3).

• pixels (torch.Tensor) – The image points. The associated pixel. The shape has to be (…, N, 2).

• intrinsics (torch.Tensor) – intrinsic matrices. The shape has to be (…, 3, 3).

• extrinsics (LieTensor) – The camera extrinsics. The shape has to be (…, 7).

Returns:

Per-pixel reprojection error. The shape is (…, N).

Example

>>> import torch, pypose as pp
>>> f, (H, W) = 2, (9, 9) # focal length and image height, width
>>> intrinsics = torch.tensor([[f, 0, H / 2],
...                            [0, f, W / 2],
...                            [0, 0,   1  ]])
>>> object = torch.randn(6, 3)
>>> pose = pp.randn_SE3()
>>> pixels = pp.point2pixel(object, intrinsics, pose)
>>> err = pp.reprojerr(object, pixels, intrinsics, pose)
>>> err
tensor([0., 0., 0., 0., 0., 0.])