Selects the face contour depending on the reletive position of the head Input: vertices: N X num_of_vertices

r"""Compute visualization outputs. Args: data (dict): Data used for the current iteration.

r"""Get inception v3 layers

r"""Initialize a dictionary of hyperparameters that we want to monitor in the HParams dashboard in tensorBoard.

r"""Every trainer should implement its own init loss function.

r"""Initialize the tensorboard. Different algorithms might require different performance metrics. Hence, custom tensorboard initializa

r"""Implement the things you want to do before saving a checkpoint. For example, you can compute the K-mean features (pix2pixHD) before

r"""Get resnet50 layers

r"""Get robust resnet50 layers

r"""Set custom debug parame.

r"""Operations to do before starting an epoch. Args: current_epoch (int): Current number of epoch.

r"""Operations to do before starting an iteration. Args: data (dict): Data used for the current iteration. current_it

r"""Get vgg16 layers

r"""Get vgg19 layers

r"""Get vgg face layers

r"""Dummy member function to be overloaded by the child class. In the child class, you can write down whatever you want to track.

r"""Write all loss values to tensorboard.

r"""Add a set of hyperparameters to be compared in tensorboard. Args: hparam_dict (dictionary): Each key-value pair in the dictionary is

vertices: [bz, nv, 3] lights: [bz, nlight, 6] returns: shading: [bz, nv, 3]

:param filepath: str :return: PIL Image

:param filepath: str :return: PIL Image

Input: images PIL list lms batch tensor

Return: transparams --numpy.array (raw_W, raw_H, scale, tx, ty) img_new --PIL.Image (target_size, target_size

Convert an angle axis to a quaternion. Adapted from ceres C++ library: ceres-solver/include/ceres/rotation.h Args: angle_axis (torch

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

same as batch_matrix2axis Calculates the rotation matrices for a batch of rotation vectors Parameters ---------- rot_vec

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

Finds the python object with the given name and calls it as a function.

d0, d1 are Variables, judge is a Tensor

Return: face_color -- torch.tensor, size (B, N, 3), range (0, 1.) Parameters: face_texture -- torc

Return: face_vertex -- torch.tensor, size (B, N, 3), in camera coordinate face_color -- torch.tensor, size (

Calculate losses, gradients, and update network weights; called in every training iteration

Return: vertex_norm -- torch.tensor, size (B, N, 3) Parameters: face_shape -- torch.tensor, size

Return: rot -- torch.tensor, size (B, 3, 3) pts @ trans_mat Parameters: angles -- tor

Return: face_shape -- torch.tensor, size (B, N, 3) Parameters: id_coeff -- torch.tensor, size

Return: face_texture -- torch.tensor, size (B, N, 3), in RGB order, range (0, 1.) Parameters: tex_coeff

1x1 convolution

1x1 convolution

for batch image Args: image (torch.Tensor): the reference tensor of shape BXHxWXC. center: [bz, 2] bboxsize: [bz, 1]

Convert a flattened parameter vector to a dictionary of parameters code_dict.keys() = ['shape', 'tex', 'exp', 'pose', 'cam', 'light']

Return: trans_m --numpy.array (2, 3) Parameters: lm --numpy.array (68, 2), y direction is opposi

NOW validation

NOW validation

Run forward pass; called by both functions <optimize_parameters> and <test>.

Run forward pass. This will be called by both functions <optimize_parameters> and <test>.

Partial fc forward, `logits = X * sample(W)`