Refer to https://github.com/tianweiy/CenterPoint L1 regression loss Args: regr (batch x max_objects x dim) gt_regr (batch

Run all_gather on arbitrary picklable data (not necessarily tensors) Args: data: any picklable object Returns: list[data]

Args: gt_boxes: (N, 8) feature_map_size: (2), [x, y] Returns:

Args: input_dict: point_features: (N1 + N2 + N3 + ..., C) batch_size: point_coord

Args: input_dict: point_features: (N1 + N2 + N3 + ..., C) batch_size: point_coord

Args: input_dict: point_features: (N1 + N2 + N3 + ..., C) batch_size: point_coord

Args: anchors: (N, 7) [x, y, z, dx, dy, dz, heading] gt_boxes: (M, 7) [x, y, z, dx, dy, dz, heading] gt_c

Class-balanced sampling of nuScenes dataset from https://arxiv.org/abs/1908.09492

Args: im: (H, W, C) [y, x] x: (N) y: (N) Returns:

Args: boxes3d_fakelidar: (N, 7) [x, y, z, w, l, h, r] in old LiDAR coordinates, z is bottom center Returns: boxes3d_lidar: [

:param boxes3d: (N, 7) [x, y, z, l, h, w, ry] in camera coords, see the definition of ry in KITTI dataset :param bottom_center: whether y is

Args: boxes_a: (N, 7) [x, y, z, dx, dy, dz, heading] boxes_b: (M, 7) [x, y, z, dx, dy, dz, heading] Returns: ans_iou

Args: boxes_a: (N, 4) [x1, y1, x2, y2] boxes_b: (M, 4) [x1, y1, x2, y2] Returns:

7 -------- 4 /| /| 6 -------- 5 . | | | | . 3 -------- 0 |/ |/ 2 -------- 1 A

Builds modules

Args: cls_scores: (N) iou_scores: (N) num_points_in_gt: (N, 7+c) cls_thresh: scalar

Calculating square distance between a and b a: [bs, c, n] b: [bs, c, m]

We assume that the 3D box has lower plane parallel to the ground. Returns: Polygon with 4 points describing the base.

fast iou algorithm. this function can be used independently to do result analysis. Must be used in CAMERA coordinate system. Args: gt_

Args: x: x.features (N, C1) out_channels: C2 Returns:

Modified from the codes of Waymo Open Dataset. Convert range images to point cloud. Args: frame: open dataset frame range

Create image depth feature volume by multiplying image features with depth distributions Args: image_features: (N, C, H,

Create dataset_infos files in order not to have it in a preprocessed pickle file with the info for each sample See PandasetDataset.get_in

Removes layers from pretrained state dict that are not used or changed in model Args: model_dict: dict, Default model sta

Finds all spconv keys that need to have weight's transposed

:param xyz: (N1 + N2 ..., 3) tensor of the xyz coordinates of the features :param xyz_batch_cnt: (batch_size), [N1, N2, ...]

Args: points: (B, N, 3) point_features: (B, N, C) boxes3d: (B, M, 7), [x, y, z, dx, dy, dz, heading]

:param input_data: input dict :return:

To support a custom dataset, implement this function to receive the predicted results from the model, and then transform the unified

Calculate average precision. Args: recalls: precisions: Returns (float): average precision. Returns: