Customized pytorch data loader collate function. Parameters ---------- batch : list / dict Batched data.

Load yaml file and return a dictionary. Parameters ---------- file : string yaml file path. opt : argparser Ar

Convert the torch tensor bounding box to o3d oabb for visualization. Parameters ---------- bbx_corner : torch.Tensor shape:

Given a list of pcd, find the minimum number and downsample all point clouds to the minimum number. Parameters ---------- pcd_np

Get projection length for the Vector projection of a onto b s.t. a_projected = length * b. (2D version) See https://en.wikipedia.org/wiki

Transformation matrix from x1 to x2. Parameters ---------- x1 : list The pose of x1 under world coordinates. x2 : list

Encode the single-channel intensity to 3 channels rgb color. Parameters ---------- intensity : np.ndarray Lidar intensity, s

Retrieve all objects in a format of (n, 7), where 7 represents x, y, z, l, w, h, yaw or x, y, z, h, w, l, yaw. Parameters

r""" Return transformation matrix for torch.affine_grid. Args: M : torch.Tensor Transformation matrix with shape :math:`(N

Remove the lidar points of the ego vehicle itself. Parameters ---------- points : np.ndarray Lidar points under lidar sensor

Remove the lidar points out of the boundary. Parameters ---------- points : np.ndarray Lidar points under lidar sensor coord

Given the index, return the corresponding data. Parameters ---------- idx : int Index given by dataloade

Convert a torch tensor to numpy. Parameters ---------- torch_tensor : torch.Tensor Returns ------- A numpy array.

r""" Transform the src based on transformation matrix M. Args: src : torch.Tensor Input feature map with shape :math:`(B,C

Data augmentation operation.

Convert the torch tensor bounding box to o3d lineset for visualization. Parameters ---------- bbx_corner : torch.Tensor shap

Calculate the average precision and recall, and save them into a txt. Parameters ---------- result_stat : dict A dictionary

r""" Return a 2-D tensor with ones on the diagonal and zeros elsewhere with the same batch size as the input. Args: n : int

The base postprocessor will generate 3d groundtruth bounding box. Parameters ---------- data_dict : dict

Generate targets for training. Parameters ---------- kwargs : list gt_box_center:(max_num, 7) R

Get disretized transformation matrix. Parameters ---------- matrix : torch.Tensor Shape -- (B, L, 4, 4) where B is the batch

Assign the attributes of lineset2 to lineset1. Parameters ---------- lineset1 : open3d.LineSet lineset2 : open3d.LineSet Re

Preprocess the lidar points to BEV representations. Parameters ---------- pcd_raw : np.ndarray The raw l

r""" Helper function to make torch.inverse work with other than fp32/64. The function torch.inverse is only implemented for fp32/64 which make

Add localization noise to the pose. Parameters ---------- pose : list x,y,z,roll,yaw,pitch xyz_

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

Denormalize the regression map Parameters ---------- reg_map : np.ndarray / torch.Tensor Regression outp

Process a single CAV's information for the train/test pipeline. Parameters ---------- selected_cav_base : dict

Get mask for the combination of cav_mask and rorated ROI mask. Parameters ---------- shape : tuple Shape of (B, L, H, W, C).

Get rorated ROI mask. Parameters ---------- shape : tuple Shape of (B,L,C,H,W). correction_matrix : torch.Tensor

Merge the preprocessed features from different cavs to the same dictionary. Parameters ---------- processed_

r""" Compute the normalization matrix from image size in pixels to [-1, 1]. Args: height : int Image height. width

Process the outputs of the model to 2D/3D bounding box. Parameters ---------- data_dict : dict The dicti

Construct bbx from the regression output of the model. Parameters ---------- reg_map : torch.Tensor Regr

Given the timestamp index, return the correct timestamp key, e.g. 2 --> '000078'. Parameters ---------- scen

Save the open3d drawing to folder. Parameters ---------- element : list List of o3d.geometry objects. save_path : str

The transformation matrix from x-coordinate system to carla world system Parameters ---------- pose : list [x, y, z, roll, y

:param input_size: (int, int) Height and width of input tensor as (height, width). :param input_dim: int e.g. 256

Return the elements in requirements.txt.

Fix the parameters of backbone during finetune on timedelay。

Fix the parameters of backbone during finetune on timedelay。

Fix the parameters of backbone during finetune on timedelay。

Fix the parameters of backbone during finetune on timedelay。

Calculate the distance to ego for each cav.

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

Collate batch if the batch is a dictionary, eg: {'voxel_features': [feature1, feature2...., feature n]} Parameters -

Collate batch if the batch is a dictionary, eg: {'voxel_features': [feature1, feature2...., feature n]} Parameters -

Customized pytorch data loader collate function. Parameters ---------- batch : dict Dict of list. Each e

Customized pytorch data loader collate function. Parameters ---------- batch : list List of dictionary.

Customized pytorch data loader collate function. Parameters ---------- batch : list List of dictionary.

Customized pytorch data loader collate function. Parameters ---------- batch : list List of dictionary.

Combine ROI mask and CAV mask Parameters ---------- roi_mask : torch.Tensor Mask for ROI region after considering the spatia

r""" Convert to homography coordinates Args: A : torch.Tensor The affine matrix with shape :math:`(B,2,3)`. Returns:

Convert 8 corners to x, y, z, dx, dy, dz, yaw. Parameters ---------- corner3d : np.ndarray (N, 8, 3) order : str

Create bounding box with 8 corners under obstacle vehicle reference. Parameters ---------- extent : list Width, height, leng

Convert the output delta to 3d bbx. Parameters ---------- deltas : torch.Tensor (N, W, L, 14) an

Convert points discretized format to continuous space for BEV representation. Parameters ---------- p_dist : numpy.array Poin

Downsample the lidar points to a certain number. Parameters ---------- pcd_np : np.ndarray The lidar points, (n, 4). nu

Given the list of the yaml files, extract the mocked timestamps. Parameters ---------- yaml_files : list

Args: data_dict: points: (N, 3 + C_in) gt_boxes: optional, (N, 7) [x, y, z, dx, dy, dz, heading]

Project the lidar and bbx to ego space first, and then do clipping. Parameters ---------- selected_cav_base : dict

Project the lidar and bbx to ego space first, and then do clipping. Parameters ---------- selected_cav_base : dict

Project the lidar and bbx to ego space first, and then do clipping. Parameters ---------- selected_cav_base : dict

Get pair-wise transformation matrix across different agents. This is only used for v2vnet and disconet. Currently we set this

Get points within a rotated bounding box (2D version). Parameters ---------- p : numpy.array Points to be tested with shape

r""" Return rotation matrix for torch.affine_grid based on transformation matrix. Args: M : torch.Tensor Transformation ma

Model inference for early fusion. Parameters ---------- batch_data : dict model : opencood.object dataset : opencood.EarlyFu

Retrieve the paths to all camera files. Parameters ---------- cav_path : str The full file path of curre