Generates sampling grid for frustum features Args: lidar_to_cam: (B, 4, 4), LiDAR to camera frame transformation

Samples input using sampling grid Args: input_features: (B, C, D, H, W), Input frustum features grid: (B, X,

Generates voxel features via 3D transformation and sampling Args: batch_dict: frustum_features: (B, C, D,

Predicts depths and creates image depth feature volume using depth distributions Args: batch_dict: images

Forward pass Args: loss: (B, H, W), Pixel-wise loss gt_boxes2d: (B, N, 4), 2D box labels for foreground/backg

Gets DDN loss Args: depth_logits: (B, D+1, H, W), Predicted depth logits depth_maps: (B, H, W), Depth map [m]

Forward pass Args: images: (N, 3, H_in, W_in), Input images Returns result: dict[torch.Tensor], Depth

Args: data_dict: spatial_features Returns:

Args: data_dict: spatial_features Returns:

Collapses voxel features to BEV via concatenation and channel reduction Args: batch_dict: voxel_features:

Args: batch_dict: encoded_spconv_tensor: sparse tensor Returns: batch_dict: s

B = batch size, S = number of cameras, C = 3, H = img height, W = img width rgb_camXs: (B,S,C,H,W) pix_T_cams: (B,S,4,4)

Args: height: (N) width: (N) min_overlap: Returns:

Gets loss Args: Returns: loss: (1), Network loss tb_dict: dict[float], All losses to log in tensorboa

Gets DDN loss Returns: loss: (1), Depth distribution network loss tb_dict: dict[float], All losses to log in

Gets DDN loss Args: Returns: loss: (1), Depth distribution network loss tb_dict: dict[float], All los

Gets number of output channels Returns: out_feature_dim: int, Number of output channels

Get padding parameters for np.pad function Args: desired_size: int, Desired padded output size cur_size: int, Current size. S

Args: voxel_coords: (N, 3) downsample_times: voxel_size: point_cloud_range: Returns:

:param u: (N) :param v: (N) :param depth_rect: (N) :return:

modified from https://github.com/open-mmlab/mmdetection Args: tcp_port: backend: Returns:

Initialize PCA whitening from given descriptors. Return tuple of shift and projection.