Implementation of the forward pass through the DPT head. This method processes a specific chunk of frames from the sequence.

Group values into groups with identical values. Args: values (torch.Tensor): [N] values to group required_group_size (in

Apply basic activation function to pose parameters. Args: pose_enc: Tensor containing encoded pose parameters act_type: Acti

Uniformly (or uniformly in inverse space) sample z values in `n_samples` bins in range [min_value, max_value]. Args: size: size of th

Calculate 3D RoPE indices for a single image. Args: image_grid_thw: Temporal, height, width dimensions of the image grid

This method is called the the thread is started, to initialize any resources that is only held in the thread.

Apply inverse log transform: sign(y) * (exp(|y|) - 1) Args: y: Input tensor Returns: Transformed tensor

Convert rotations given as rotation matrices to quaternions. Args: matrix: Rotation matrices as tensor of shape (..., 3, 3). Re

Convert a pose encoding back to camera extrinsics and intrinsics. This function performs the inverse operation of extri_intri_to_pose_encoding,

Multiply two quaternions. Usual torch rules for broadcasting apply. Args: a: Quaternions as tensor of shape (..., 4), real part

Extract the first or third Euler angle from the two members of the matrix which are positive constant times its sine and cosine. Args:

Extract the first or third Euler angle from the two members of the matrix which are positive constant times its sine and cosine. Args:

Applies 1D rotary position embeddings along one dimension. Args: tokens: Input token features. positions: Position in

Apply positional embedding to tensor x.

Ensure the value is divisible by the stride.

Creates a causal 4D mask of shape `(batch_size, 1, query_length, key_value_length)` from a 2D mask of shape `(batch_size, key_value_l

Preprocess an image or batch of images. Copy of the `preprocess` method from `CLIPImageProcessor`. Args: images (`ImageI

Preprocess an image or batch of images. Copy of the `preprocess` method from `CLIPImageProcessor`. Args: images (`ImageI

Applies Rotary Position Embedding to the query and key tensors. Args: q (`torch.Tensor`): The query tensor. k (`torch.Tensor`): T

Converts a NumPy array to a PIL image. Automatically: - Squeezes dimensions of size 1. - Moves the channel dimension (if it has s

r"""Sample a tensor using bilinear interpolation `bilinear_sampler(input, coords)` samples a tensor :attr:`input` at coordinates :attr:`coord

Calculate the Area Under the Curve (AUC) for the given error arrays. :param r_error: numpy array representing R error values (Degree). :

Computes the inverse of each 4x4 SE3 matrix in the batch. This function assumes PyTorch3D coordinate. Args: - se3 (Tensor): Nx4x4 t

Compute face normals of a triangular mesh Args: vertices (torch.Tensor): [..., N, 3] 3-dimensional vertices faces (torch.Ten

Compute face normals of a triangular mesh Args: vertices (np.ndarray): [..., N, 3] 3-dimensional vertices faces (np.ndarray)

This function: - first downsizes the image with LANCZOS inteprolation, which is better than bilinear interpolation in

Custom interpolate to avoid INT_MAX issues in nn.functional.interpolate.

Evaluate the depth map using various metrics and return a depth error parity map, with an option for least squares alignment. Args:

Convert a depth map to world coordinates. Args: depth_map (np.ndarray): Depth map of shape (H, W). intrinsic (np.ndarray): C

DINOv2 ViT-L/14 model with registers (optionally) pretrained on the LVD-142M dataset.