A counter for the `thop` package to count the operations in an attention operation. Meant to be used like: macs, params = thop.pr

Create folder for data preprocessing.

Decode bboxes. Args: cls_scores (Tensor): Outputs from the classification head, \ shape [num_query, cls_out_channe

Initialize from serialized content.

3x3 convolution with padding

Extract features of images.

Extract features of images.

Applies filtering to boxes. Distance, bike-racks and points per box. :param nusc: An instance of the NuScenes class. :param eval_boxes: A

Forward function' Args: pts_feats (list[torch.Tensor]): Features of point cloud branch gt_bboxes_3d (list[:obj:`BaseIn

Forward training function. Args: points (list[torch.Tensor], optional): Points of each sample. Defaults to None.

Generate one 2D annotation record given various informations on top of the 2D bounding box coordinates. Args: ann_rec (dict): Origina

Generate one 2D annotation record given various informations on top of the 2D bounding box coordinates. Args: ann_rec (dict): Origina

Get the 2D annotation records for a given `sample_data_token`. Args: sample_data_token (str): Sample data token belonging to a camera \

Get attribute from predicted index. This is a workaround to predict attribute when the predicted velocity is not reliable. We map the

Convert instance masks to axis-aligned bounding boxes. Args: points (np.array): Scene points of shape (n, 6). pts_ins

Get semantic segmentation map for an image. Args: nuim (obj:`NuImages`): NuImages dataset object img_info (dict): Meta informatio

Get data infos. This method gets information from the raw data. Args: num_workers (int): Number of threads to be used. D

Returns the data path as well as all annotations related to that sample_data. Note that the boxes are transformed into the current sensor's

Compute scene_idxs for data sampling and label weight for loss \ calculation. We sample more times for scenes with more points. Label

Compute scene_idxs for data sampling and label weight for loss \ calculation. We sample more times for scenes with more points. Label

Compute regression and classification targets for a batch image. Outputs from a single decoder layer of a single feature level are used.

Waymo annotation format version like KITTI: { [optional]points: [N, 3+] point cloud [optional, for kitti]image: {

Convert the box from global to camera coordinate. Args: info (dict): Info for a specific sample data, including the calibratio

Parameters ---------- xy: Tensor xy coordinates. shape=(N, ..., 2) E.g., (N, 2) or (N, K, 2) or (N, H, W, 2) Returns

Initialize layers of the Detr3DTransformer.

Default initialization for Parameters of Module.

Default initialization for Parameters of Module.

Initialize the transformer weights.

Default initialization for Parameters of Module.

Returns a new Pose that corresponds to the inverse of this one. Returns ---------- result: Pose Inverted

Prepare data related to Kitti dataset. Related data consists of '.pkl' files recording basic infos, 2D annotations and groundtruth database.

Loads ground truth boxes from DB. :param nusc: A NuScenes instance. :param eval_split: The evaluation split for which we load GT boxes.

Loss function for outputs from a single decoder layer of a single feature level. Args: cls_scores (Tensor): Box score logi

Convert keys in checkpoints for VoteNet. There can be some breaking changes during the development of mmdetection3d, and this tool is used fo

Convert rotations given as rotation matrices to quaternions. Args: matrix: Rotation matrices as tensor of shape (..., 3, 3). Re

Convert boxes from :obj:`NuScenesBox` to :obj:`CameraInstance3DBoxes`. Args: boxes (list[:obj:`NuScenesBox`]): List of predicted NuScenesB