Iteratively refine camera pose predictions. Args: pose_tokens (torch.Tensor): Normalized camera tokens with shape [B, 1,

https://gist.github.com/bbrighttaer/207dc03b178bbd0fef8d1c0c1390d4be

Upsample flow field [H/P, W/P, 2] -> [H, W, 2] using convex combination

Args: windows: (num_windows*B, window_size, window_size, C) window_size (int): Window size H (int): Height of image

Build the patch embed layer. If 'conv', we use a simple PatchEmbed conv layer. Otherwise, we use a vision transformer.

r""" The call function to the pipeline for generation. Args: image (`PipelineImageInput`): The input imag

r""" The call function to the pipeline for generation. Args: image (`PipelineImageInput`): The input imag

Function invoked when calling the pipeline for generation. Args: image (`PipelineImageInput`): The input

Function invoked when calling the pipeline for generation. Args: image (`PipelineImageInput`): The input

Perform the attention computation, integrating image features (if provided) and timestep embeddings. If `ip_hidden_states` is `None`

Generates spatial positions for a batch of patches. Args: batch_size: Number of samples in the batch. height: Height

Args: dataset_dict (dict): Metadata of one image, in Detectron2 Dataset format. Returns: dict: a format that

Args: dataset_dict (dict): Metadata of one image, in Detectron2 Dataset format. Returns: dict: a format that

Args: dataset_dict (dict): Metadata of one image, in Detectron2 Dataset format. Returns: dict: a format that

Args: original_image (np.ndarray): an image of shape (H, W, C) (in BGR order). Returns: predictions (dict):