Convert a pose encoding back to camera extrinsics and intrinsics. This function performs the inverse operation of extri_intri_to_pose_encoding,

Preprocess image tensor(s) to target size with crop or pad mode. Args: img_tensor (torch.Tensor): Image tensor of shape (C, H, W) or

Args: rgb (bool): whether to return RGB colors or BGR colors. maximum (int): either 255 or 1 Returns: ndarray: a vect

r"""Sample spatial features `sample_features4d(input, coords)` samples the spatial features :attr:`input` represented by a 4D tensor :math:`(

r""" Sets the attention processor to use to compute attention. Parameters: processor (`dict` of `AttentionProcessor` or o

Processes specialized tokens with shape (1, 2, X, C) for multi-frame processing: 1) Uses the first position (index=0) for the first frame onl

Solve `min |focal * xy / (z + shift) - uv|` with respect to shift and focal

Solve `min |focal * xy / (z + shift) - uv|` with respect to shift

Compute and display summary metrics for evaluation results given a custom value for max_dets_per_image

Compute and display summary metrics for evaluation results given a custom value for max_dets_per_image

args: rgbs: [S C H W] R: [S 3 3] T: [S 3] xy_depth: [S H W 3] focal_length: [S] pick_idx: list of

Args: x: (B, H, W, C) window_size (int): window size Returns: windows: (num_windows*B, window_size, window_size, C)

Args: num_frames (int): If this is none, it means that we need find gen frames num by ourselves test_num_frames (int)

Args: region_target_height (int): Height without the padding, the padding will be scaled together region_target_w

Args: region_target_height (int): Height without the padding, the padding will be scaled together region_target_w

Args: region_target_height (int): Height without the padding, the padding will be scaled together region_target_w

split the video into several overlapped windows. args: x: the input video frames. [B, T, C, H, W] outp

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

Initializes the position generator with an empty cache.

Build a pyramid of feature maps from the input. fmaps: Tensor (B, S, C, H, W) num_levels: number of pyramid levels (each dow

observed_depth: float

Args: cfg (CfgNode): num_gpus (int): if 0, will run on CPU

Computes frequency components for rotary embeddings. Args: dim: Feature dimension (must be even). seq_len: Maximum se

Convert different format of masks or polygons to a tuple of masks and polygons. Returns: list[GenericMask]:

Derive the desired score numbers from summarized COCOeval. Args: coco_eval (None or COCOEval): None represents no predic

Find proper places to draw text given a binary mask.

Evaluate the box proposals in predictions. Fill self._results with the metrics for "box_proposals" task.

Evaluate predictions. Fill self._results with the metrics of the tasks.

Evaluate predictions. Fill self._results with the metrics of the tasks.

Evaluate detection proposal recall metrics. This function is a much faster alternative to the official COCO API recall evaluation code. Howev

Evaluate the coco results using COCOEval API.

RoPE for image tokens with 2d structure. Args: embed_dim: (`int`): The embedding dimension size crops_coords (`Tuple[int]`)

r""" Creates 3D sinusoidal positional embeddings. Args: embed_dim (`int`): The embedding dimension of inputs. It must be

Process frame attention blocks. We keep tokens in shape (B*S, P, C).

Process frame attention blocks. We keep tokens in shape (B*S, P, C).

Process global attention blocks. We keep tokens in shape (B, S*P, C).

Process global attention blocks. We keep tokens in shape (B, S*P, C).

Performs feature rotation by splitting and recombining feature dimensions. Args: x: Input tensor to rotate. Returns:

Args: Same as in :meth:`__init__()`. Returns: fig (matplotlib.pyplot.figure): top level container for all the