Copy `master_params` to `model_params`.

Copy the `model_params` gradients to `master_params` for the optimizer step.

Conert pandaset pose dict to a numpy vector in order to pass it through the network

Args: batch_dict: batch_size: batch_cls_preds: (B, num_boxes, num_classes | 1) or (N1+N2+..., num

Preprocess images Args: images: (N, 3, H, W), Input images Return x: (N, 3, H, W), Preprocessed image

Only validate in KITTIDataset Args: gt_boxes: (N, 7 + C) [x, y, z, dx, dy, dz, heading, ...] road_planes: [a,

Calculate the yaw angle from a quaternion. Note that this only works for a quaternion that represents a box in lidar or global coordinate fra

Calculate the yaw angle from a quaternion. Note that this only works for a quaternion that represents a box in lidar or global coordinate fra

Read in a calibration file and parse into a dictionary.

Read the values inside the optimizer for the hyper-parameters.

Args: batch_dict: batch_size: rois: (B, num_rois, 7 + C) point_coords: (num_point

Args: batch_dict: batch_size: rois: (B, num_rois, 7 + C) point_coords: (num_point

Args: batch_dict: batch_size: rois: (B, num_rois, 7 + C) spatial_features_2d: (B,

Args: batch_dict: batch_size: rois: (B, num_rois, 7 + C) point_coords: (num_point

Args: batch_dict: batch_size: rois: (B, num_rois, 7 + C) point_coords: (num_point

Applies a rotation. :param rot_matrix: <np.float: 3, 3>. Rotation matrix. :return: <None>.

Args: points: (B, N, 3 + C) angle: (B), angle along z-axis, angle increases x ==> y Returns:

Args: batch_dict: batch_size: rois: (B, num_rois, 7 + C) roi_scores: (B, num_rois

Args: class_name: sample_group: Returns:

Args: points: (N, 3) num_sampled_points: int num_sectors: int Returns: sampled_points: (N_out, 3)

Args: roi_boxes: (M, 7 + C) points: (N, 3) Returns: sampled_points: (N_out, 3)

Returns: Serialized instance as dict.

PyTorch Implementation for tf.nn.sigmoid_cross_entropy_with_logits: max(x, 0) - x * z + log(1 + exp(-abs(x))) in https://www.

Applies a homogeneous transform. :param transf_matrix: <np.float: 4, 4>. Homogenous transformation matrix. :return: <None>.

Use transformer to fuse feature. template_feature : BxCxN template_xyz : BxNx3

Transforms voxel sampling grid into frustum sampling grid Args: grid: (B, X, Y, Z, 3), Voxel sampling grid gr

Applies a translation to the point cloud. :param x: <np.float: 3, 1>. Translation in x, y, z. :return: <None>.

4-------- 6 /| /| 5 -------- 3 . | | | | . 7 -------- 1 |/ |/

Args: xyz: (N, 3) xyz_batch_cnt: (batch_size) features: (N, C) new_xyz: (M, 3) ne

Args: data_dict: gt_boxes: (N, 7 + C) [x, y, z, dx, dy, dz, heading, ...] Returns:

To support a custom dataset, implement this function to load the raw data (and labels), then transform them to the unified normative

Class for manipulating and viewing point clouds. :param points: <np.float: 4, n>. Input point cloud matrix.

:param center: [<float>: 3]. Center of box given as x, y, z. :param size: [<float>: 3]. Size of box in width, length, height.