Get the valid radios of feature maps of all level.

Interpolate the `source` to the shape of the `target`. The `source` must be a Tensor, but the `target` can be a Tensor or a np.ndarray with t

Make divisible function. This function rounds the channel number to the nearest value that can be divisible by the divisor. It is taken from

Parse raw annotation to target format. Args: raw_data_info (dict): Raw data information load from ``ann_file`` Returns:

Draw a confusion matrix by matplotlib. Modified from `Scikit-Learn <https://github.com/scikit-learn/scikit-learn/blob/dc580a8ef/sklea

Perform forward propagation of the tracking head and predict tracking results on the features of the upstream network. Args:

Run Kalman filter prediction step. Args: mean (ndarray): The 8 dimensional mean vector of the object state at the

Perform forward propagation of the bbox head and predict detection results on the features of the upstream network. Args:

Simple version of ``kwimage.Boxes.random`` Returns: Tensor: shape (n, 4) in x1, y1, x2, y2 format. References: https://gitlab.

Makes a function retry itself after encountering pytorch's CUDA OOM error. The implementation logic is referred to https://gi

Set ``gt_label``.

Generate grid anchors of a single level. Note: This function is usually called by method ``self.grid_priors``. Args:

Make sure different ranks share the same seed. All workers must call this function, otherwise it will deadlock. This method is generally used

See :func:`BaseInstanceMasks.to_ndarray`.

Box head forward function used in both training and testing. Args: stage (int): The current stage in Cascade RoI Head.

Dispatch kwargs to preprocess(), forward(), visualize() and postprocess() according to the actual demands. Returns: Tuple

Return an activation function given a string.

Post-process the output of retriever. Args: result (torch.Tensor): Score matrix of single retrieve, either from i

A context which ignores CUDA OOM exception from pytorch. Code is modified from <https://github.com/facebookresearch/detectron2/blob/main/dete

Preprocess the inputs to a list. Cast the input data to a list of data. - list or tuple: return inputs - str: -

Mask head forward function used in both training and testing. Args: x (tuple[Tensor]): Tuple of multi-level img features.

A `to_tuple` function generator. It returns a function, this function will repeat the input to a tuple of length ``n`` if the input is not an

Multi-stage refinement of RoI. Args: x (tuple[Tensor]): List of multi-level img features. rois (Tensor): shape (n, 5)

Display or save image with groung truths and predictions from a model. Args: dataset (Dataset): A PyTorch dataset.

Set the range of magnitudes.

Get the flag indicate whether anchor centers are inside regions.

Assign gt to priors. Args: pred_instances (:obj:`InstanceData`): Instances of model predictions. It includes ``pr

Assign gt to approxs. This method assign a gt bbox to each group of approxs (bboxes), each group of approxs is represent by a base ap

Assign w.r.t. the overlaps of priors with gts. Args: overlaps (Tensor): Overlaps between k gt_bboxes and n bboxes,

Pooling the feature tokens. Args: x (torch.Tensor): The input tensor, should be with shape ``(B, num_heads, L, C)`` or ``(B,

Decode bounding box based on distances. Args: points (Tensor): Shape (n, 2) or (b, n, 2), [x, y]. bbox (Tensor): Shape (n, 4) or

Map bboxes from testing scale to original image scale.

Check the queues' state.

We load parameter values from ema model to source model before validation. Args: runner (Runner): The runner of the train

Calculate a proportional bbox region. The bbox center are fixed and the new h' and w' is h * ratio and w * ratio. Args: bbox (Tensor

Calculate region of the box defined by the ratio, the ratio is from the center of the box to every edge.

Clamp the values of a vector to the range [-MAX_CLAMP_VALUE, MAX_CLAMP_VALUE]. Args: vector (Tensor): Tensor of shape (N, C, H, W).

This function converts target class indices to one-hot vectors, given the number of classes. Args: targets (Tensor): The ground truth

Convert coordinate tensor to positional encoding. Args: coord_tensor (Tensor): Coordinate tensor to be converted to positiona

Crop the images according to some bounding boxes. Typically for re- identification sub-module. Args: img (Tensor): of sha

The normalization layer of decoder.

Decode distance prediction to bounding box. Args: points (Tensor): Shape (B, N, 2) or (N, 2). distance (Tensor): Distance from th

Get the dtype.

Calculate recalls. Args: gts (list[ndarray]): a list of arrays of shape (n, 4) proposals (list[ndarray]): a list of arrays of sha

extract the appropriate t index for a batch of indices.

Create fake boxes with specific sizes and fill values. Args: sizes (Tuple[int]): The size of fake boxes. The last value must

Filter pipeline to avoid KeyError with partial data info.

Flip boxes horizontally or vertically in-place. Args: img_shape (Tuple[int, int]): A tuple of image height and width.

Forward features from the upstream network. Args: x (tuple[Tensor]): Features from the upstream network, each is

Generate proposals from encoded memory. The function will only be used when `as_two_stage` is `True`. Args: memory (Tenso

Generate proposals from encoded memory. Args: memory (Tensor) : The output of encoder, has shape (bs, num_key, em

Get custom accuracy w.r.t. cls_score and labels. Args: cls_score (Tensor): The prediction with shape (N, C + 2). labe

Get the certain type boxes from masks. Please refer to ``mmdet.structures.bbox.box_type`` for more details of the box type.

Get the width and height of boxes with type of tensor or box type. Args: boxes (Tensor or :obj:`BaseBoxes`): boxes with type of tensor

Get length of one video. Args: idx (int): Index of video. Returns: int (int): The length of the video.

Args: norm (str or callable): either one of BN, SyncBN, FrozenBN, GN; or a callable that takes a channel number and returns

Get positive information from sampling results. Returns: list[:obj:`InstanceData`]: Positive Information of each image,

Get ``num_points`` most uncertain points with random points during train. Sample points in [0, 1] x [0, 1] coordinate space based on their

Create track from unassociated measurement. Args: measurement (ndarray): Bounding box coordinates (x, y, a, h) with

Check if the pytorch version is lower than "version.

extract the last associated observation.

Map from level to magnitude.

Load annotations from an annotation file named as ``self.ann_file`` Returns: List[dict]: A list of annotation.

load detection boxes.