pose: [N, 17*3]

Flip the flipped heatmaps back to the original form. Note: - batch_size: N - num_keypoints: K - heatmap height: H

x_0: joint gaussian noise X_T x_1: twist gaussian noise X_T

Sample features following the index. Note: B: batch size N: point number C: channel number of each point Ns: sampl

Initialize model weights.

Make divisible function. This function rounds the channel number down to the nearest value that can be divisible by the divisor. Args:

CPU version of multi-scale deformable attention. Args: value (Tensor): The value has shape (bs, num_keys, mum_heads, embed_dim

保持检测框中心点位置不变调整检测框比例为cfg.input_img_size ration: 扩大检测框大小

Freeze parameters.

Freeze parameters.

Freeze parameters.

Modulate heatmap distribution with Gaussian. sigma = 0.3*((kernel_size-1)*0.5-1)+0.8 sigma~=3 if k=17 sigma=2 if k=11; sigma~=1.5

To calculate the affine matrix, three pairs of points are required. This function is used to get the 3rd point, given 2D points a & b. The 3r

Check if the pose results contain track_id.

Load checkpoint from somewhere (modelzoo, file, url). Args: filename (str): Accept local filepath, URL, ``torchvision://xxx``,

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

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

Converts axis-angle representation to 6D rotation representation. Args: axis_angle: Tensor of shape [bs, 3], where each row is a

Applies a batch of rigid transformations to the joints Parameters ---------- rot_mats : torch.tensor BxNx3x3 Tensor of rotat

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

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

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

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

Convert axis-angle representation to rotation matrix. Args: theta: size = [B, 3] Returns: Rotation matrix corresponding to the

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

Calculates the per vertex displacement due to the blend shapes Parameters ---------- betas : torch.tensor Bx(num_betas) Blend s

Calculates the per vertex displacement due to the blend shapes Parameters ---------- betas : torch.tensor Bx(num_betas) Blend s

Build filters function.

Computes a similarity transform (sR, t) that takes a set of 3D points S1 (3 x N) closest to a set of 3D points S2, where R is an 3x3 rota

Computes a similarity transform (sR, t) that takes a set of 3D points S1 (3 x N) closest to a set of 3D points S2, where R is an 3x3 rota

Compute the twist component of given rotation and twist axis https://stackoverflow.com/questions/3684269/component-of-a-quaternion-rotation-a

3x3 convolution with padding

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)

Decode the keypoints from output regression. Args: img_metas (list(dict)): Information about data augmentation By

If a key contains 'keypoints', and f'{key}_mask' is in self.keys(), invalid zeros will be inserted to the right places and f'{key}_mask'

Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks). This is the same as the DropConnect impl I created for Ef

Forward function.

Get the area of a bbox_xyxy. Args: (Union[list, tuple]): A list of [x1, y1, x2, y2]. bbox_convention (str, optional):

Get the expansion of a residual block. The block expansion will be obtained by the following order: 1. If ``expansion`` is given, just retur

For every initial grid, get its index in the feature map. Note: [H_init, W_init]: shape of initial grid [H, W]: shape of feature m

neighbour_matrix: 最终的邻接矩阵，表示所有关节和边之间的邻接关系 neighbour_matrix[num_joints:, num_joints:]: 仅表示边之间的邻接关系的子矩阵 neighbour_m