Copy all files from src_dir to dst_dir that match any of the given suffixes

Encodes the input by passing through the encoder network and returns the latent codes. :param input: (Tensor) Input tensor to

Get estimated bones from prior shape and corresponding bone feature and encoding

Get estimated bones from prior shape and corresponding bone feature and encoding

Save all vertices in clip space project to screen, with visibility mask (..., 4) KT evaluation will map source gt keypoint to the nearest vis

Input posed bones in world space and mvp matrix, output keypoint coordinates in clip space head -> tail -> LF -> RF -> LR -> RR Following

Create new Mesh class with the n-th (included) mesh to the m-th (not included) mesh in the batch. Args: m: the index of

Create new Mesh class with only the n-th meshes in the batch. Args: n: the index of the mesh to be contained. R

Return difference of two masks, where pixels are 1 if different, 0 if same

Plot metrics. `stats` should be a dictionary of type stats[dataset][t][metric][i] where dataset is the dataset name (e.g. `train` or `

Reparameterization trick to sample from N(mu, var) from N(0,1). :param mu: (Tensor) Mean of the latent Gaussian [B x D]

Save model, optimizer, and metrics state to a checkpoint in checkpoint_dir for the specified iteration.