Reload new_zeros from self.tensor.

Cancatenates a box instance list into one single box instance. Similar to ``torch.cat``. Args: box_list (Sequence[T]): A

Reload ``reshape`` from self.tensor.

Reload ``view`` from self.tensor.

Update the tracker. Args: kwargs (dict[str: Tensor | int]): The `str` indicates the name of the input variable. `

Reload ``permute`` from self.tensor.

Reload ``unsqueeze`` from self.tensor.

Get the buffer of a specific item. Args: item (str): The demanded item. ids (list[int], optional): The demanded ids.

Concatenates a sequence of tensors along a new dimension. Similar to ``torch.stack``. Args: box_list (Sequence[T]): A seq

Build normalization layer. Args: cfg (dict): The norm layer config, which should contain: - type (str): Layer type.

Reload ``to`` from self.tensor.

Reload ``repeat`` from self.tensor.

Reload ``flatten`` from self.tensor.

Reload ``transpose`` from self.tensor.

Reload ``split`` from self.tensor.

Get the model's metainfo by the model name. Args: model_name (str): The name of model. Returns: modelindex.m

Reload ``new_zeros`` from self.tensor.

Reload ``numpy`` from self.tensor.

Reload ``detach`` from self.tensor.

Reload ``squeeze`` from self.tensor.

Reload ``new_tensor`` from self.tensor.

Whether the buffer is empty or not.

Reload ``expand`` from self.tensor.

Reload ``dim`` from self.tensor.

Change the device of the data inplace. Example: >>> self = SamplingResult.random() >>> print(f'self = {self.to(None)}

Reload ``numel`` from self.tensor.

dict: a dictionary of info about the object

Reload ``cpu`` from self.tensor.

Reload ``clone`` from self.tensor.

Apply function to a list of arguments. Note: This function applies the ``func`` to multiple inputs and map the multiple outputs o

Normalization.

Decode the predicted bboxes according to prediction and base boxes.

Reload ``new_full`` from self.tensor.

Print a log message.

Create a superset of inputs needed to run test or train batches. Args: batch_size (int): batch size. Defaults to 2. image_shapes

Loading information for evaluation. Args: record (dict): Label information or test results. The format might look

Reload ``new_ones`` from self.tensor.

Convert targets by image to targets by feature level. [target_img0, target_img1] -> [target_level0, target_level1, ...]

Overrides this method to set the :attr:`device` Returns: nn.Module: The model itself.

Grab configs necessary to create a detector. These are deep copied to allow for safe modification of parameters without influencing other tes

Reload ``cuda`` from self.tensor.

Convert a list of bboxes to roi format. Args: bbox_list (List[Union[Tensor, :obj:`BaseBoxes`]): a list of bboxes correspondin

Obtain the mean of tensor on different GPUs.

Check whether the ``pipeline_results`` is the same with the predefined ``results``. Args: results (dict): Predefined results which sh

Generate random bitmap masks for demo / testing purposes. Example: >>> from mmdet.data_elements.mask.structures import BitmapMask

Assign gt to priors. The assignment is done in following steps 1. compute iou between all prior (prior of all pyramid levels) and gt

Unmap a subset of item (data) back to the original set of items (of size count)

Sample positive and negative bboxes. This is a simple implementation of bbox sampling given candidates, assigning results and ground

Run per image evaluation on given images and store results (a list of dict) in self.evalVids :return: None

Register all modules in mmdet into the registries. Args: init_default_scope (bool): Whether initialize the mmdet default scope.

Integrate test results.

Get tensor data from box type boxes. Args: boxes (Tensor or BaseBoxes): boxes with type of tensor or box type. If its type is

Convert bbox coordinates from (cx, cy, w, h) to (x1, y1, x2, y2). Args: bbox (Tensor): Shape (n, 4) for bboxes. Returns: Ten

Calculate overlap between two set of bboxes. FP16 Contributed by https://github.com/open-mmlab/mmdetection/pull/4889 Note: Assume bbo

See :func:`BaseInstanceMasks.pad`.

nn.Module: the normalization layer named "norm1"

Create a list of fake porposals. Args: image_shapes (list[tuple[int]]): Batch image shapes. num_proposals (int): The number of fa

Convert the model into training mode will keeping the normalization layer freezed.

Unpack ``gt_instances``, ``gt_instances_ignore`` and ``img_metas`` based on ``batch_data_samples`` Args: batch_data_samples (List[:ob

nn.Module: the normalization layer named "norm2"

Since teacher_model is registered as a plain object, it is necessary to put the teacher model to other device when calling ``to`` func

Draw datasample and save to all backends. - If GT and prediction are plotted at the same time, they are displayed in a stitched image

Handle predicted instances when RoI is empty. Note: If ``instance_results`` is not None, it will be modified in place internally, and then re

Inverse function of sigmoid. Args: x (Tensor): The tensor to do the inverse. eps (float): EPS avoid numerical overflow. Defaults

Convert bbox coordinates from (x1, y1, x2, y2) to (cx, cy, w, h). Args: bbox (Tensor): Shape (n, 4) for bboxes. Returns: Ten

Get anchors according to feature map sizes. Args: featmap_sizes (list[tuple]): Multi-level feature map sizes. batch_i

Generate grid anchors in multiple feature levels. Args: featmap_sizes (list[tuple]): List of feature map sizes in

See :func:`BaseInstanceMasks.flip`.

Create a superset of inputs needed to run test or train batches. Args: batch_size (int): batch size. Default to 1. num_frames (in

Extract features from images.

Resize pos_embed weights. Args: pos_embed (torch.Tensor): Position embedding weights with shape [1, L, C]. src_shape

Compute losses of the head. Args: cls_scores (list[Tensor]): Cls and quality scores for each scale level has shap

Encode deltas between bboxes and ground truth boxes.

benchmark entry method. Args: repeat_num (int): Number of repeat benchmark. Defaults to 1.

Extract a multi-scale single image tensor from a multi-scale batch tensor based on batch index. Note: The default value of detach is True, be

Grab configs necessary to create a roi_head. These are deep copied to allow for safe modification of parameters without influencing other tes

Convert objects of various python types to :obj:`torch.Tensor`. Supported types are: :class:`numpy.ndarray`, :class:`torch.Tensor`, :class:`S

Encode the input images and get corresponding results.

Calculate the loss based on the features extracted by the DIIHead. Args: cls_score (Tensor): Classification prediction

Geometric transformation for bbox. Args: bboxes (Union[torch.Tensor, np.ndarray]): Shape (n, 4) for bboxes. homography_matrix (Un

Get targets for PAA head. This method is almost the same as `AnchorHead.get_targets()`. We direct return the results from _get_target

Transform a batch of output features extracted from the head into bbox results. This method is same as `BaseDenseHead.get_results()`.

See :func:`BaseInstanceMasks.to_tensor`.

Apply element-wise weight and reduce loss. Args: loss (Tensor): Element-wise loss. weight (Optional[Tensor], optional): Element-w