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 query (N, Length_{query}, C) :param reference_points (N, Length_{query}, n_levels, 2), range

Forward Function of MultiScaleDeformAttention. Args: query (Tensor): Query of Transformer with shape ( bs, num_que

Uses iterative farthest point sampling to select a set of npoint features that have the largest minimum distance :param ctx:

:param ctx: :param features: (B, C, N) :param idx: (B, npoint) index tensor of the features to gather :return:

Find the three nearest neighbors of unknown in known :param ctx: :param unknown: (B, N, 3) :param known: (B, M, 3)

Performs weight linear interpolation on 3 features :param ctx: :param features: (B, C, M) Features descriptors to be interpol

:param ctx: :param features: (B, C, N) tensor of features to group :param idx: (B, npoint, nsample) tensor containing the ind

:param ctx: :param radius: float, radius of the balls :param nsample: int, maximum number of features in the balls :p

:param xyz: (B, N, 3) xyz coordinates of the features :param new_xyz: (B, npoint, 3) centroids :param features: (B, C, N) des

:param xyz: (B, N, 3) xyz coordinates of the features :param new_xyz: ignored :param features: (B, C, N) descriptors of the f

:param xyz: (B, N, 3) tensor of the xyz coordinates of the features :param features: (B, N, C) tensor of the descriptors of the the f

:param unknown: (B, n, 3) tensor of the xyz positions of the unknown features :param known: (B, m, 3) tensor of the xyz positions of

Args: ctx: max_range: int, max range of voxels to be grouped nsample: int, maximum number of features in

Args: new_coords: (M1 + M2 ..., 3) centers voxel indices of the ball query xyz: (N1 + N2 ..., 3) xyz coordinates of t

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

Args: ctx: radius: float, radius of the balls nsample: int, maximum number of features in the balls

Args: ctx: features: (N1 + N2 ..., C) tensor of features to group features_batch_cnt: (batch_size) [N1 +

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

Args: ctx: xyz: (B, N, 3) where N > npoint npoint: int, number of features in the sampled set Re

Args: ctx: xyz: (N1 + N2 + ..., 3) where N > npoint xyz_batch_cnt: [N1, N2, ...] npoint: int,

Args: ctx: unknown: (N1 + N2..., 3) unknown_batch_cnt: (batch_size), [N1, N2, ...] known: (M1

Args: ctx: features: (M1 + M2 ..., C) idx: [N1 + N2 ..., 3] weight: [N1 + N2 ..., 3]

Args: ctx: // support_xyz: (N1 + N2 ..., 3) xyz coordinates of the features // xyz_batch_cnt: (batch_size

Args: ctx: support_xyz: (N1 + N2 ..., 3) xyz coordinates of the features xyz_batch_cnt: (batch_size), [N1

Args: unknown: (N1 + N2 ..., 3) known: (M1 + M2 ..., 3) unknow_feats: (N1 + N2 ..., C1) known

Args: support_xyz: (N1 + N2 ..., 3) xyz coordinates of the features support_features: (N1 + N2 ..., C) point-wise fea

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

Args: input: (B, #anchors, #classes) float tensor. Predicted logits for each class target: (B, #ancho

Args: input: (B, #anchors, #codes) float tensor. Ecoded predicted locations of objects. target: (B, #

Args: input: (B, #anchors, #codes) float tensor. Ecoded predicted locations of objects. target: (B, #

Args: input: (B, #anchors, #classes) float tensor. Predited logits for each class. target: (B, #ancho

Args: output: (batch x dim x h x w) or (batch x max_objects) mask: (batch x max_objects) ind: (batch x ma

Args: input: (B, #anchors, #classes) float tensor. Predicted logits for each class target: (B, #ancho

Args: input: (B, #anchors, #codes) float tensor. Ecoded predicted locations of objects. target: (B, #

Args: input: (B, #anchors, #codes) float tensor. Ecoded predicted locations of objects. target: (B, #

Args: input: (B, #anchors, #classes) float tensor. Predited logits for each class. target: (B, #ancho

Args: output: (batch x dim x h x w) or (batch x max_objects) mask: (batch x max_objects) ind: (batch x ma

query: (B, N, C)

query: (B, N, C) key: (S, M, B, C) reference_points_cam: (S, B, N, D, 2), in 0-1 bev_mask: (S, B, N, D)

Input x, shape (N, in_channels, H_ori, W_ori)

Input x, shape (N, in_channels, X_ori, Y_ori, Z_ori)

Input x, shape (N, C, H, W)

Input x, shape (N, C, X, Y, Z)

Applies convolutional block Args: features: (B, C_in, H, W), Input features Returns: x: (B, C_out, H,

Args: batch_dict: batch_size: int vfe_features: (num_voxels, C) voxel_coords: (nu

Args: batch_dict: batch_size: int vfe_features: (num_voxels, C) voxel_coords: (nu

Args: batch_dict: batch_size: int vfe_features: (num_voxels, C) voxel_coords: (nu

Args: batch_dict: batch_size: int vfe_features: (num_voxels, C) points: (num_poin

Args: batch_dict: batch_size: int vfe_features: (num_voxels, C) points: (num_poin

Args: batch_dict: batch_size: int vfe_features: (num_voxels, C) voxel_coords: (nu

Args: batch_dict: batch_size: keypoints: (B, num_keypoints, 3) multi_scale_3d_fea

Applies convolutional block Args: features: (B, C_in, H, W), Input features Returns: x: (B, C_out, H,

Applies convolutional block Args: features: (B, C_in, H, W), Input features Returns: x: (B, C_out, H,

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

Predicts depths and creates image depth feature volume using depth distributions Args: images: (N, 3, H_in, W_in), Input

Args: batch_dict: images: (N, 3, H_in, W_in), Input images **kwargs: Returns: bat

Args: batch_dict: voxels: (num_voxels, max_points_per_voxel, C) voxel_num_points: optional (num_v

Args: voxels: (num_voxels, max_points_per_voxel, C) voxel_num_points: optional (num_voxels) Returns:

Args: **kwargs: Returns: batch_dict: ... vfe_features: (num_voxels, C)