Import the module "models/[model_name].py Parameters __________ hypes : dict Dictionary containing parameters. Returns

:param self: :param input_tensor: (b, c, h, w) input is actually the target_model :param h_cur: (b, c_hidden, h,

:param input_tensor: (b, t, c, h, w) or (t,b,c,h,w) depends on if batch first or not extracted features from alexnet :par

Parameters ---------- output_dict : dict target_dict : dict

Compute loss for pixor network Parameters ---------- output_dict : dict The dictionary that contains the o

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

Parameters ---------- output_dict : dict target_dict : dict

Generate targets for training. Parameters ---------- argv : list gt_box_center:(max_num, 7), anchor:(H,

Generate mask to remove the bounding boxes outside the range. Parameters ---------- boxes : torch.Tensor Groundtruth bbx

Get points within a rotated bounding box (3D version). Parameters ---------- p : numpy.array Points to be tested with shape

Args: gt_boxes: (N, 7 + C), [x, y, z, dx, dy, dz, heading, [vx], [vy]] points: (M, 3 + C), rot_range: [min, max] Retu

Args: gt_boxes: (N, 7), [x, y, z, dx, dy, dz, heading] points: (M, 3 + C), scale_range: [min, max] Returns:

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

Model inference for late fusion. Parameters ---------- batch_data : dict model : opencood.object dataset : opencood.LateFusi

Given the extrinsic matrix, project lidar data to another space. Parameters ---------- lidar_data : np.ndarray Lidar data, s

Associate two lists of lineset. Parameters ---------- vis : open3d.Visualizer lineset_list1 : list lineset_list2 : list

Load bev related geometry parameters s.t. boundary, resolutions, input shape, target shape etc. Parameters ---------- param : di

Based on the lidar range and resolution of voxel, calcuate the anchor box and target resolution. Parameters ---------- param : d

Load saved model if exiseted Parameters __________ saved_path : str model saved path model : opencood object The

Based on the lidar range and resolution of voxel, calcuate the anchor box and target resolution. Parameters ---------- param : d

Based on the lidar range and resolution of voxel, calcuate the anchor box and target resolution. Parameters ---------- param : d

Print out the loss function for current iteration. Parameters ---------- epoch : int Current epoch for

Print out the loss function for current iteration. Parameters ---------- epoch : int Current epoch for

Apply non-maximum suppression to avoid detecting too many overlapping bounding boxes for a given object. Parameters ---------- b

Performs rorated non-maximum suppression and returns indices of kept boxes. Parameters ---------- boxes : torch.tensor The lo

Read pcd and return numpy array. Parameters ---------- pcd_file : str The pcd file that contains the point cloud. Retu

Process the outputs of the model to 2D bounding box. Step1: convert each cav's output to bounding box format Step2: project t

Process the outputs of the model to 2D/3D bounding box. Step1: convert each cav's output to bounding box format Step2: projec

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

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

Preprocess the lidar points by simple sampling. Parameters ---------- pcd_np : np.ndarray The raw lidar.

Preprocess the lidar points by voxelization. Parameters ---------- pcd_np : np.ndarray The raw lidar.

Project the 3d bounding box to another coordinate system based on the transfomration matrix. Parameters ---------- box3d : torch

Project the points to another coordinate system based on the transfomration matrix. Parameters ---------- points : torch.Tensor

Project points to BEV occupancy map with default ratio=0.1. Parameters ---------- points : np.ndarray (N

Project the objects under world coordinates into another coordinate based on the provided extrinsic. Parameters ---------- objec

Stack all projected lidar together. Parameters ---------- projected_lidar_list : list The list containing all projected lida

Args: gt_boxes: (N, 7 + C), [x, y, z, dx, dy, dz, heading, [vx], [vy]] points: (M, 3 + C) Returns:

Args: gt_boxes: (N, 7 + C), [x, y, z, dx, dy, dz, heading, [vx], [vy]] points: (M, 3 + C) Returns:

Remove bounding box that has negative z axis. Parameters ---------- bbx_3d : torch.Tensor Predcited 3d bounding box, shape:(

Avoid adding ego vehicle to the object dictionary. Parameters ---------- objects : dict The dictionary contained all objects

Remove large bounding box. Parameters ---------- bbx_3d : torch.Tensor Predcited 3d bounding box, shape:(N,8,3) Returns

Retrieve the ego vehicle id from sample(origin format). Parameters ---------- base_data_dict : dict Data sample in origin fo

Args: points: (B, N, 3 + C) angle: (B), radians, angle along z-axis, angle increases x ==> y Returns:

Rorate the points along z-axis. Parameters ---------- points : torch.Tensor / np.ndarray (N, 2). angle : torch.Tensor / n

Save prediction and gt tensor to txt file.

Save the dictionary into a yaml file. Parameters ---------- data : dict The dictionary contains all data. save_name : s

Set up the learning rate schedular. Parameters ---------- hypes : dict The training configurations. optimizer : torch.o

Create optimizer corresponding to the yaml file Parameters ---------- hypes : dict The training configurations. model :

Create folder for saved model based on current timestep and model name Parameters ---------- hypes: dict Config yaml diction

Visualize the prediction, ground truth with point cloud together. Parameters ---------- pred_box_tensor : torch.Tens

Visualize a frame during inference for video stream. Parameters ---------- pred_box_tensor : torch.Tensor (N, 8, 3) predicti

Visualize the batch data in animation. Parameters ---------- dataloader : torch.Dataloader Pytorch dataloader order : s

Visualize the prediction, groundtruth with point cloud together in a bev format. Parameters ---------- pred_box : torch.Tensor

Visualize the prediction, groundtruth with point cloud together. Parameters ---------- pred_tensor : torch.Tensor (N, 8, 3)