Args: dataset_cfg: EasyDict() with key POINT_CLOUD_RANGE POINT_FEATURE_ENCODING D

init will set values for: self.class_names self.cls2label (after mapping) self.label2cls (after mapping)

Args: ap_iou_thresh: float between 0 and 1.0 IoU threshold to judge whether a prediction is positive.

Compute 3D bounding box IoU. Input: corners1: numpy array (8,3), assume up direction is negative Y corners2: numpy array (8,3),

Use accumulated predictions and groundtruths to compute Average Precision. # 1. calculate by-category precision, recall with multiprocessing m

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

Quantize point cloud by voxel_size Returns kept indices Args: all_points: np.array (n, 3) float voxel_sz: float Returns:

In the forward pass we receive a Tensor containing the input and return a Tensor containing the output. ctx is a context object that

Rotation points w.r.t. rotmat Args: pc: np.array (n, 3) rotmat: np.array (4, 4) Returns: pc: (n, 3)

Args: im: (m, n)

Generate pose matrix with z-dim as height Args: pose: np.array (4, 4) Returns: urc: (4, 4) urc_inv: (4, 4)