Convert a list of bboxes to roi format. Args: bbox_list (list[Tensor]): a list of bboxes corresponding to a batch of images.

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

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

Decode the predicted bboxes according to prediction and base boxes.

See :func:`BaseInstanceMasks.expand`.

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

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

See :func:`BaseInstanceMasks.to_ndarray`.

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

Build loss.

Args: size (tuple): expected shape of dummy masks, (H, W) or (N, H, W) Return: ndarray: dummy mask

Args: size (tuple): expected shape of dummy masks, (N, H, W) Return: list[list[ndarray]]: dummy mask

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

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

Initialize the weights of FPN module.

Compute losses of the head. Args: cls_scores (list[Tensor]): Box scores for each scale level Has shape (N, num_an

Create random AssignResult for tests or debugging. Args: num_preds: number of predicted boxes num_gts: number of true

See :func:`BaseInstanceMasks.resize`.

Build detector.

Recursively convert Tensor in inputs from src_type to dst_type. Args: inputs: Inputs that to be casted. src_type (torch.dtype): S

Inference image(s) with the detector. Args: model (nn.Module): The loaded detector. imgs (str/ndarray or list[str/ndarray]): Eith

Build head.

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

Calculate accuracy according to the prediction and target. Args: pred (torch.Tensor): The model prediction, shape (N, num_class)

Builder of box sampler.

Check if is ResNet building block.

NMS for multi-class bboxes. Args: multi_bboxes (Tensor): shape (n, #class*4) or (n, 4) multi_scores (Tensor): shape (n, #class),

Calculate the ious between each bbox of bboxes1 and bboxes2. Args: bboxes1(ndarray): shape (n, 4) bboxes2(ndarray): shape (k, 4)

Builder of box assigner.

Mask head forward function used in both training and testing.

Convert detection results to a list of numpy arrays. Args: bboxes (Tensor): shape (n, 5) labels (Tensor): shape (n, ) num

Get segmentation masks from mask_pred and bboxes. Args: mask_pred (Tensor or ndarray): shape (n, #class, h, w). F

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

Assign gt to bboxes. This method assign a gt bbox to every bbox (proposal/anchor), each bbox will be assigned with -1, or a semi-posi

Encode deltas between bboxes and ground truth boxes.

See :func:`BaseInstanceMasks.flip`.

Get targets for GFL head. This method is almost the same as `AnchorHead.get_targets()`. Besides returning the targets as the parent m

Initialize a detector from config file. Args: config (str or :obj:`mmcv.Config`): Config file path or the config object.

Convert masks to the format of ndarray.

Map bboxes from the original image scale to testing scale.

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

See :func:`BaseInstanceMasks.pad`.

Create a superset of inputs needed to run test or train batches. Args: input_shape (tuple): input batch dimensions n

Build a module. Args: cfg (dict, list[dict]): The config of modules, is is either a dict or a list of configs. regist

Get root logger. Args: log_file (str, optional): File path of log. Defaults to None. log_level (int, optional): The level of logg

Parse the package dependencies listed in a requirements file but strips specific versioning information. Args: fname (str): path to r

Apply element-wise weight and reduce loss. Args: loss (Tensor): Element-wise loss. weight (Tensor): Element-wise weights.

Box forward function used in both training and testing.

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

Initialize conv sequential for CornerHead.

Parse the raw outputs (losses) of the network. Args: losses (dict): Raw output of the network, which usually contain

Transpose and gather feature according to index. Args: feat (Tensor): Target feature map. ind (Tensor): Target coord

Build roi extractor.

See :func:`BaseInstanceMasks.crop`.

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

Merge augmented mask prediction. Args: aug_masks (list[ndarray]): shape (n, #class, h, w) img_shapes (list[ndarray]): shape (3, )

nn.Module: the normalization layer named "norm1"

Check whether the anchors are inside the border. Args: flat_anchors (torch.Tensor): Flatten anchors, shape (n, 4). valid_flags (t

Builder of box coder.

Build PyTorch DataLoader. In distributed training, each GPU/process has a dataloader. In non-distributed training, there is only one dataload

Decode distance prediction to bounding box. Args: points (Tensor): Shape (n, 2), [x, y]. distance (Tensor): Distance from the giv

Evaluate in VOC protocol. Args: results (list[list | tuple]): Testing results of the dataset. metric (str | list[str]

Get category ids of concatenated dataset by index. Args: idx (int): Index of data. Returns: list[int]: All c

Inference image(s) with the detector. Args: model (nn.Module): The loaded detector. imgs (str/ndarray or list[str/ndarray]): Eith

dict: a dictionary of info about the object

Regress the bbox for the predicted class. Used in Cascade R-CNN. Args: rois (Tensor): shape (n, 4) or (n, 5) label (T

see :func:`BaseInstanceMasks.resize`

Calculate and round number of filters based on depth multiplier.

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

Args: num_channels: conv_channels: first_time: whether the input comes directly from the efficientnet,

Box head forward function used in both training and testing.

Box head forward function used in both training and testing.

Build backbone.

Builder of IoU calculator.

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

Encode bitmap mask to RLE code. Args: mask_results (list | tuple[list]): bitmap mask results. In mask scoring rcnn, mask_resu

Directly extract features from the backbone+neck.

Transform network output for a batch into bbox predictions. Args: cls_scores (list[Tensor]): Box scores for each scale level