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

Format the results to json (standard format for COCO evaluation). Args: results (list[tuple | numpy.ndarray]): Testing results o

Placeholder of forward function.

Forward training function. Args: multi_level_feats (list[torch.Tensor]): Multi level img_feats. voxel_semantics (to

Forward training function. Args: img_metas (list[dict[dict]]): Meta information of each sample.

Forward training function. Args: img_metas (list[dict[dict]]): Meta information of each sample.

Fuse long term history into current BEV. Args: curr_bev (torch.Tensor): Current BEV. shape (B, base_bev_channels, H, W)

Args: features (torch.Tensor): (len-1, c, z, h, w) locations (torch.Tensor): (len-1, w, h, z, 3) returns:

Use grid sample to gather feature from previous BEV volume. Args: features (torch.Tensor): shape (bs, len, w, h, z, c).

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

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

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

get the annotation info according to the index. Args: index (Int): the index of the data. Returns: an

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

Get available scenes from the input nuscenes class. Given the raw data, get the information of available scenes for further info generation.

Get available scenes from the input nuscenes class. Given the raw data, get the information of available scenes for further info generation.

This method is used to obtain Bird's Eye View (BEV) features from multi-level features, BEV queries, and other related parameters. us

obtain bev features.

Get reference point for reshaped BEV volume. Args: W (int): width of the BEV volume. H (int): height of the

Get reference point for reshaped BEV volume. Args: W (int): width of the BEV volume. H (int): height of the

Get data info according to the given index. Args: index (int): Index of the sample data to get. Returns: dic

Get file names of waymo raw data.

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

sample the input index list Args: idx_list (List[int]): the index list from `index - self.history_len` to `index`.

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

Calculate refine_feat_loss from refine_feat_w Args: refine_feat_w (torch.Tensor): The weight without sigmoid. shape (b

Calculate refine_feat_loss from refine_feat_w Args: refine_feat_w (torch.Tensor): The weight without sigmoid. shape (b

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

Args: img_metas (List[dict]): Meta information of each sample. length = batch_size. With ke

obtain bev features.

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 calibra

This matrix is called by the function `compute_mIoU`. But I move the count matrix process to the `forward_test` function in the mo

Pre-compute the necessary information in acceleration including the index of points in the final feature. Args: coor (tor

Initialize layers of the Detr3DTransformer.

Initialize layers of the Detr3DTransformer.

Initialize layers of the Detr3DTransformer.

Initialize layers of the Detr3DTransformer.

Initialize layers of the Detr3DTransformer.

Initialize layers of the Detr3DTransformer.

Default initialization for Parameters of Module.

Default initialization for Parameters of Module.

Default initialization for Parameters of Module.

Default initialization for Parameters of Module.

Initialize the transformer weights.

Default initialization for Parameters of Module.

Default initialization for Parameters of Module.

For convenience, assumes the sampling dimension is 0. This also fills in the ends with 0 Args: x: the :math:`x`-coordinates at w

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

Load annotations from ann_file. Args: ann_file (str): Path of the annotation file. Returns: list[dict

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

Args: voxel_semantics (torch.Tensor): occ_outs (torch.Tensor): Predicted occupancy features with shape (bs, w, h,

Compute segmentation loss.

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