Loop to ``full_init`` each dataset.

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

Fuse the parameters in a branch with a conv and bn. Args: conv (nn.Conv2d): The convolution module to fuse. bn (nn.BatchNorm2d):

r"""Generate 2D gaussian radius. This function is modified from the `official github repo <https://github.com/princeton-vl/CornerNet-Lite/blo

Generate base anchors. Returns: list(torch.Tensor): Base anchors of a feature grid in multiple \ feature levels.

Base on the previous bboxes and regression values, we compute the regressed bboxes and generate the grids on the bboxes. Args:

Generate base anchors of a single level. Args: base_size (int | float): Basic size of an anchor. scales (torch.Tensor

embed_dim: output dimension for each position pos: a list of positions to be encoded: size (M,) out: (M, D)

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

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

Get `gt_group_of` of a certain class, which is used in Open Images. Args: annotations (list[dict]): Same as `eval_map()`. class_i

Get targets in denoising part for a batch of images. Args: batch_gt_instances (list[:obj:`InstanceData`]): Batch of

Get the reference points used in encoder. Args: spatial_shapes (Tensor): Spatial shapes of features in all levels,

Get gt_masks originally or generated based on bboxes. If gt_masks is not contained in results, it will be generated based on gt_bboxe

Get the number of ignore bboxes.

Convert bytes to int.

Get the match score. Args: bboxes (torch.Tensor): of shape (num_current_bboxes, 4) in [tl_x, tl_y, br_x, br_y] fo

Get a pre-defined model or create a model from config. Args: model (str | Config): The name of model, the config file path or a

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

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

Get the position embedding of proposal.

Get relative positional embeddings according to the relative positions of query and key sizes. Borrowed from https://github.com/facebookresea

Get relative positional embeddings according to the relative positions of query and key sizes. Args: q_size (int): size of query

Get sampled approxs and inside flags according to feature map sizes. Args: featmap_sizes (list[tuple]): Multi-level feature map s

Make dataset by walking all images under a root. Args: root (string): root directory of folders folder_to_idx (dict): the map fro

Compute corresponding GT box and classification targets for proposals. Args: proposals_list (list[Tensor]): Multi level p

Generate corner targets. Including corner heatmap, corner offset. Optional: corner embedding, corner guiding shift, centripetal shif

Calculate the ground truth for all samples in a batch according to the sampling_results. Args: sampling_results (List[obj

Calculate the ground truth for all samples in a batch according to the sampling_results.". Args: sampling_results (List[:

Calculate the ground truth for all samples in a batch according to the sampling_results. Almost the same as the implementation in bbo

generate sine position embedding from a position tensor Args: pos_tensor (torch.Tensor): shape: [..., n]. num_pos_feats (int): pro

Get the tokens positive and prompts for the caption. Args: original_caption (str): The original caption, e.g. 'bench . car .'

Returns the index of the transform in a pipeline. Args: pipeline (List[dict] | List[BaseTransform]): The transforms list. target

Estimate uncertainty based on pred logits. We estimate uncertainty as L1 distance between 0.0 and the logits prediction in 'mask_preds' for t

Get a tensor of bboxes centered at (0, 0). Returns: Tensor: Tensor of bboxes with shape (num_bboxes, 4) in [xmin, ymi

Return nested attribute of obj Example: getattr_recursive(obj, 'a.b.c') is equivalent to obj.a.b.c

Convert ground truth to coco format json file. Args: gt_dicts (Sequence[dict]): Ground truth of the dataset. outfile_

Convert ground truth to coco format json file. Args: gt_dicts (Sequence[dict]): Ground truth of the dataset. outfile_

Convert box coordinates from (x1, y1, x2, y2) to corners ((x1, y1), (x2, y1), (x1, y2), (x2, y2)). Args: boxes (Tensor):

Re-normalize the image. Args: img (Tensor | ndarray): Input image. If the input is a Tensor, the shape is (1, C, H, W). If th

Instance segmengation postprocess. Args: mask_cls (Tensor): Classfication outputs of shape (num_queries, cls_out_

Are points located in bboxes. Args: points (Tensor): Points, shape: (m, 2). bboxes (Tensor): Bounding boxes, shape: (n, 4).

Randomly jitter the given color in order to better distinguish instances with the same class. Args: color (tuple): The RGB color tupl

return the observation k step away before.

Maximization part of EM algorithm(Expectation-Maximization)

L2 loss. Args: pred (torch.Tensor): The prediction. target (torch.Tensor): The learning target of the prediction. Returns:

Load anns with the specified ids. ``self.anns`` is a list of annotation lists instead of a list of annotations. Args:

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

Log image size. Args: img_scale (tuple): Image size to be logged. shape_order (str, optional): The order of image shape.

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

Calculate the loss in a batch. Args: bbox_feats (Tensor): of shape [N, C, H, W]. N is the number of bboxes.

Perform forward propagation and loss calculation of the detection roi on the features of the upstream network. Args: x (t

Args: batch_inputs (Tensor): Input images of shape (N, C, H, W). These should usually be mean centered and std sc

Calculate losses from multi-branch inputs and data samples. Args: multi_batch_inputs (Dict[str, Tensor]): The dict of multi-branc

generate train_loss from the input tensor and data_samples. Args: images (Tensor): A batch of images. The shape of it should be

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 detection head. Args: cls_scores (list[Tensor]): Box scores for

Calculate the loss based on the features extracted by the mask head. Args: mlvl_kernel_preds (list[Tensor]): Multi-level dynamic

Calculate the loss based on the features extracted by the detection head. Args: cls_scores (list[Tensor]): Box scores for

Calculate the loss based on the features extracted by the detection head. Args: cls_scores (list[Tensor]): Box scores for

Calculate the loss based on the features extracted by the detection head. Args: cls_scores (list[Tensor]): Box scores for

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

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

Compute losses of the head. Args: center_heatmap_preds (list[Tensor]): center predict heatmaps for all levels with

Calculate the loss based on the features extracted by the detection head. Args: pred_maps (list[Tensor]): Prediction map

Calculate the loss based on the features extracted by the detection head. Args: cls_scores (list[Tensor]): Box scores for

Calculate the loss based on the features extracted by the mask head. Args: mask_preds (list[Tensor]): List of predicted masks, ea

Compute losses of the head. Args: anchor_list (list[list[Tensor]]): Multi level anchors of each image.

loss_by_feat() is implemented in StageCascadeRPNHead.

Calculate the loss based on the features extracted by the detection head. Args: cls_scores (list[Tensor]): Box scores for

Calculate the loss based on the features extracted by the detection head. Args: cls_scores (list[Tensor]): Box scores for

Calculate the loss based on the features extracted by the detection head. Args: cls_scores (list[Tensor]): Box scores for

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

Calculate the loss based on the features extracted by the detection head. Args: cls_scores (list[Tensor]): Box scores for

Calculate the loss based on the features extracted by the detection head. Args: cls_scores (list[Tensor]): Box scores for

Calculate the loss based on the features extracted by the detection head. Args: tl_heats (list[Tensor]): Top-left corner

Calculate the loss based on the features extracted by the mask head. Args: mask_preds (list[Tensor]): List of predicted masks, ea

Calculate the loss based on the features extracted by the detection head. Args: cls_scores (list[Tensor]): Cls and qualit

Calculate the loss based on the features extracted by the detection head. Args: cls_scores (list[Tensor]): Box scores for

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

Calculate the loss based on the features extracted by the detection head. Args: cls_scores (list[Tensor]): Box scores for

Loss function. Only outputs from the last feature level are used for computing losses by default. Args: all_laye

Calculate denoising loss. Args: all_layers_denoising_cls_scores (Tensor): Classification scores of all decoder la