Get the normalization layer. Args: norm_layer (str): The type of normalization layer. Returns: norm_layer (nn.Module):

Generate transformation matrix.

Convert a single ijkl one-hot array to rotation direction string. Mapping: [1, 0, 0, 0] (i) -> 'up' [0, 1, 0, 0] (j) ->

Initialize a inference state.

r""" Initialize model parameters using Xavier initialization.

r""" Initialize model parameters using Xavier initialization.

r""" Initialize model parameters using Xavier initialization.

r""" Initialize model parameters using Xavier initialization.

Get final keypoint predictions from heatmaps and transform them back to the image. Note: - batch size: N - num keypoints: K

features: shape=[1, T, 512] input_fps: fps for audio, f_a output_fps: fps for video, f_m output_len: video length

input 133x3 numpy keypoints and output AAPoseMeta Args: kp2ds (List[np.ndarray]): _description_ width (int): _descrip

Load images from various input types. Args: inputs (Union[str, np.ndarray, List[np.ndarray]]): Input can be file

Load the video frames from a directory of JPEG files ("<frame_index>.jpg" format). The frames are resized to image_size x image_size and are

Load the video frames from a directory of JPEG files ("<frame_index>.jpg" format). The frames are resized to image_size x image_size and are

One step for the second-order multistep DPMSolver. Args: model_output_list (`List[torch.Tensor]`): The di

One step for the third-order multistep DPMSolver. Args: model_output_list (`List[torch.Tensor]`): The dir

One step for the UniC (B(h) version). Args: this_model_output (`torch.Tensor`): The model outputs at `x_

One step for the UniP (B(h) version). Alternatively, `self.solver_p` is used if is specified. Args: model_output (`torch

Normalize commas and remove all whitespace from user input.

Return the smallest value >= frame_num that satisfies F = 4n + 1.

DARK post-pocessing. Implemented by udp. Paper ref: Huang et al. The Devil is in the Details: Delving into Unbiased Data Processing for Human

Performs post-processing on the model's output to extract bounding boxes, scores, and class IDs. Args: input_image (nump

Opt-in pre-warm. Run one dummy DiT forward at the same shape a subsequent generate() call will use, so CUDA kernels are autotuned, FSD

Convert rotation direction strings to one-hot encoded array. Mapping: 'up' (i) -> [1, 0, 0, 0] 'left' (j) -> [0, 1, 0, 0

Build (F,4) WASD and IJKL float arrays (0/1) from parsed segments.

Predict the sample from the previous timestep by reversing the SDE. This function propagates the sample with the multistep DPMSolver.

Predict the sample from the previous timestep by reversing the SDE. This function propagates the sample with the multistep UniPC.

r""" Reconstruct video tensors from patch embeddings. Args: x (List[Tensor]): List of patchified features

r""" Reconstruct video tensors from patch embeddings. Args: x (List[Tensor]): List of patchified features

Reconstruct video tensors from patch embeddings. Args: x (List[Tensor]): List of patchified features, ea

Process input and estimate 2D keypoints. Args: inputs (Union[str, np.ndarray, List[np.ndarray]]): Input can be f

r""" Initializes the image-to-video generation model components. Args: config (EasyDict): Object containi

r""" Initializes the image-to-video generation model components. Args: config (EasyDict): Object containi