check points is in 3d convex polygons. Args: points: [num_points, 3] array. polygon_surfaces: [num_polygon, max_num_surfaces,

convert kitti locations, dimensions and angles to corners. format: center(xy), dims(xy), angles(clockwise when positive) Args:

convert kitti locations, dimensions and angles to corners Args: centers (float array, shape=[N, 3]): locations in kitti label file

Finds the CUDA install path.

Create optimizer learning rate scheduler based on config. Args: learning_rate_config: A LearningRate proto message. Returns: A learning

BatchNorm layers to have parameters in single precision. Find all layers and convert them back to float. This can't be done w

generate relative box corners based on length per dim and origin point. Args: dims (float array, shape=[N, ndim]): array of leng

Kitti eval. Only support 2d/bev/3d/aos eval for now. Args: gt_annos: dict, must from get_label_annos() in kitti_common.py dt_annos

Kitti eval. support 2d/bev/3d/aos eval. support 0.5:0.05:0.95 coco AP. Args: gt_annos: dict, must from get_label_annos() in kitti_common.p

Create boolean mask by actually number of a padded tensor. Args: actual_num ([type]): [description] max_num ([type]): [descriptio

Creates dense vector with indices set to specific value and rest to zeros. This function exists because it is unclear if it is safe to use tf.s

Compute pairwise intersection areas between boxes. Args: boxes1: a numpy array with shape [N, 4] holding N boxes boxes2: a nu

convert point cloud to voxels, create targets if ground truths exists.

Unmap a subset of item (data) back to the original set of items (of size count)

Pillar Feature Net Layer. The Pillar Feature Net could be composed of a series of these layers, but the PointPillars paper results on

Computes area of boxes. Args: boxes: Numpy array with shape [N, 4] holding N boxes Returns: a numpy array with shape [

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

convert 3d box corners from corner function above to surfaces that normal vectors all direct to internal. Args: corners (float ar

generate relative box corners based on length per dim and origin point. Args: dims (float array, shape=[N, ndim]): array of

flat a nested json-like dict. this function make shadow copy.