Get subset of serialized data information list. Args: indices (int or Sequence[int]): If type of indices is int,

Serialize ``self.data_list`` to save memory when launching multiple workers in data loading. This function will be called in ``full_init``.

sum over the first and last dimention

Spatial Relative Positional Embeddings.

Spatial Relative Positional Embeddings.

Spatial Relative Positional Embeddings.

Spatial Relative Positional Embeddings.

Contrast distortion. Args: img (np.ndarray): The input image. Returns: np.ndarray: Image after contrast chang

Encode images with backbone and decode into a semantic segmentation map of the same size as input.

Execute an replication callback `__data_parallel_replicate__` on each module created by original replication. The callback will be invoked w

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

Load annotation file and set ``BaseDataset._fully_initialized`` to True. If ``lazy_init=False``, ``full_init`` will be called during

Get DIOR category ids by index. Args: idx (int): Index of data. Returns: List[int]: All categories in the im

Get category ids by index. Dataset wrapped by ClassBalancedDataset must implement this method. The ``ClassBalancedDataset`` requires

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

Perform forward propagation of the head, then calculate loss and predictions from the features and data samples. Args: x

Perform forward propagation of the head, then calculate loss and predictions from the features and data samples. Args: x

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

Calculate the loss based on the features extracted by the mask head. Args: mask_preds (Tensor): Predicted foreground masks, has s

Calculate the loss based on the features extracted by the detection head.

Perform forward propagation and loss calculation of the mask head on the features of the upstream network. Args: x (tuple

Parse raw annotation to target format. This method should return dict or list of dict. Each dict or list contains the data informatio

Transform a batch of output features extracted from the head into bbox results. Note: When score_factors is not None, the cls_scores

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

Get data processed by ``self.pipeline``. Args: idx (int): The index of ``data_info``. Returns: Any: Depends

Random select some elements from the gallery. If `gallery` is a Tensor, the returned indices will be a Tensor; If `gallery` is a ndar

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

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

Convert ``xyxy`` style bounding boxes to ``xywh`` style for COCO evaluation. Args: bbox (numpy.ndarray): The bounding box

Args: master_callback: a callback to be invoked after having collected messages from slave devices.

Run forward function and calculate loss for auxiliary head in training.

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

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

bbox post-processing method. The boxes would be rescaled to the original image scale and do the nms operation. Usually `with_nms` is

bbox post-processing method. The boxes would be rescaled to the original image scale and do the nms operation. Args:

bbox post-processing method. The boxes would be rescaled to the original image scale and do the nms operation. Args:

Compute the mean and standard-deviation with sum and square-sum. This method also maintains the moving average on the master device.

Deepcopy for all attributes other than ``data_list``, ``data_address`` and ``data_bytes``. Args: memo: Memory dict which

Function to randomly crop images, bounding boxes, masks, semantic segmentation maps. Args: results (dict): Result dict fr

Run forward function and calculate loss for decode head in training.

Paste instance masks according to boxes. This implementation is modified from https://github.com/facebookresearch/detectron2/ Args:

Network forward process. Usually includes backbone, neck and head forward without any post- processing.