Cauculate mean per joint position error (MPJPE) or its variants P-MPJPE. Args: keypoint_results (list): Keypoint predicti

Resize and concatenate the feature_maps. Args: feature_maps (list[Tensor]): Feature maps. align_corners (bool): Align corners whe

Send output of the process method to registered output buffers. Args: output_msg (Message): output message force (boo

Transform inputs for decoder. Args: inputs (list[Tensor] | Tensor): multi-level img features. Returns: Tenso

Generate the target heatmap via 'UDP' approach. Paper ref: Huang et al. The Devil is in the Details: Delving into Unbiased Data Processing

Transform the joints from their world coordinates to their camera coordinates. Note: - N: number of joints Args:

remap v/f and vn/fn to vt/ft. Args: vmapping (np.ndarray, optional): the mapping relationship from f to ft. Defaults to None.

Check if the weight(and bias) is all zero.

query_points: [N, 3]

auto calculate the vertex normals.

auto calculate the uv coordinates. Args: cache_path (str, optional): path to save/load the uv cache as a npz file, this can avoid

Convert rotations given as Euler angles in radians to axis/angle. Args: axis_angle: Rotations given as a vector in axis angle form,

Convert rotations given as axis/angle to rotation matrices. Args: axis_angle: Rotations given as a vector in axis angle form,

Calculates the rotation matrices for a batch of rotation vectors Parameters ---------- rot_vecs: torch.tensor Nx3 array of N axis-

Calculates the rotation matrices for a batch of rotation vectors Parameters ---------- rot_vecs: torch.tensor Nx3

RT: (N, 3, 4) intrinsics: (N, 3, 2), [[fx, fy], [cx, cy], [width, height]]

RT: (N, 3, 4) intrinsics: (N, 3, 2), [[fx, fy], [cx, cy], [width, height]]

Builds the closure for the edge objective

Method for creating a model from a path and a model type Parameters ---------- model_path: str Either the path to the model you w

Builds the closure for the vertex objective

Returns list of utf-8 byte and a corresponding list of unicode strings. The reversible bpe codes work on unicode strings. This means you

Return the output dimension

Check if all elements in target_keys is in result_keys.

Check if norm layer is in correct train state.

Check if norm layer is in correct train state.

Check if norm layer is in correct train state.

Check if norm layer is in correct train state.

Contrast distortion.

Convert applicable model parameters to low-precision (bf16 or fp16)

Copy the image region and paste to the background. Args: img (np.ndarray): Image data. background_img (np.ndarray): Background im

Creates a directory if not exist.

Criterion of wingloss. Note: batch_size: N num_keypoints: K Args: pred (torch.Tensor[NxKxHxW]):

Criterion of wingloss. Note: batch_size: N num_keypoints: K dimension of keypoints: D (D=2 or D=3)

Crops positional embeddings for SD3 compatibility.

Decompose the standard extrinsic matrix into RT. Batched I/O.

Code modified from https://github.com/shubham-goel/4D-Humans/blob/a0def798c7eac811a63c8220fcc22d983b39785e/hmr2/models/components/pose_transformer.py#

Prediction each pixel.

Prediction each pixel.