Data Parallel with a replication callback. An replication callback `__data_parallel_replicate__` of each module will be invoked after being

Convert video array to Tensor.

Produce a keypoint representation that will be further used by other modules

Dataset of videos, videos can be represented as an image of concatenated frames, or in '.mp4','.gif' format

Hourglass Decoder

Hourglass Encoder

Hourglass architecture.

Detecting a keypoints. Return keypoint position and variance.

Motion transfer generator. That Given a keypoints and an appearance trying to reconstruct the target frame. Produce 2 versions of target fram

Randomly change the brightness, contrast and saturation and hue of the clip Args: brightness (float): How much to jitter brightness. brightnes

Module that predicting a dense optical flow only from the displacement of a keypoints and the appearance of the first frame

Discriminator similar to Pix2Pix

Merge all generator related updates into single model for better multi-gpu usage

Simple block for processing video (encoder).

Simple block for processing video (encoder).

A thread-safe future implementation. Used only as one-to-one pipe.

Merge all generator related updates into single model for better multi-gpu usage

Dataset of detected keypoints

Dataset of pairs for transfer.

RNN for predicting kp movement

Extract random crop at the same location for a list of images Args: size (sequence or int): Desired output size for the crop in format (h,

Resizes a list of (H x W x C) numpy.ndarray to the final size The larger the original image is, the more times it takes to interpolate Arg

Rotate entire clip randomly by a random angle within given bounds Args: degrees (sequence or int): Range of degrees to select from If

Res block, preserve spatial resolution.

Simple block with group convolution.

Pipe for master-slave communication.

An abstract `SyncMaster` object. - During the replication, as the data parallel will trigger an callback of each module, all slave devices should

Simple block for processing video (decoder).

r"""Applies Synchronized Batch Normalization over a 2d or 3d input that is seen as a mini-batch. .. math:: y = \frac{x - mean[x]}{ \

r"""Applies Batch Normalization over a 4d input that is seen as a mini-batch of 3d inputs .. math:: y = \frac{x - mean[x]}{ \sqrt{Va

r"""Applies Batch Normalization over a 5d input that is seen as a mini-batch of 4d inputs .. math:: y = \frac{x - mean[x]}{ \sqrt{Va