from typing import ValuesView import torch.nn as nn import torch import numpy as np import copy import torch.nn.functional as F from pcdet.ops.iou3d_nms import iou3d_nms_utils from ...utils import common_utils, loss_utils from .roi_head_template import RoIHeadTemplate from ..model_utils.mppnet_utils import build_transformer, PointNet, MLP from .target_assigner.proposal_target_layer import ProposalTargetLayer from pcdet.ops.pointnet2.pointnet2_stack import pointnet2_modules as pointnet2_stack_modules class MPPNetHeadE2E(RoIHeadTemplate): def __init__(self,model_cfg, num_class=1,**kwargs): super().__init__(num_class=num_class, model_cfg=model_cfg) self.model_cfg = model_cfg self.use_time_stamp = self.model_cfg.get('USE_TIMESTAMP',None) self.num_lidar_points = self.model_cfg.Transformer.num_lidar_points self.avg_stage1_score = self.model_cfg.get('AVG_STAGE1_SCORE', None) self.nhead = model_cfg.Transformer.nheads self.num_enc_layer = model_cfg.Transformer.enc_layers hidden_dim = model_cfg.TRANS_INPUT self.hidden_dim = model_cfg.TRANS_INPUT self.num_groups = model_cfg.Transformer.num_groups self.grid_size = model_cfg.ROI_GRID_POOL.GRID_SIZE self.num_proxy_points = model_cfg.Transformer.num_proxy_points self.seqboxembed = PointNet(8,model_cfg=self.model_cfg) self.jointembed = MLP(self.hidden_dim*(self.num_groups+1), model_cfg.Transformer.hidden_dim, self.box_coder.code_size * self.num_class, 4) num_radius = len(self.model_cfg.ROI_GRID_POOL.POOL_RADIUS) self.up_dimension_geometry = MLP(input_dim = 29, hidden_dim = 64, output_dim =hidden_dim//num_radius, num_layers = 3) self.up_dimension_motion = MLP(input_dim = 30, hidden_dim = 64, output_dim = hidden_dim, num_layers = 3) self.transformer = build_transformer(model_cfg.Transformer) self.roi_grid_pool_layer = pointnet2_stack_modules.StackSAModuleMSG( radii=self.model_cfg.ROI_GRID_POOL.POOL_RADIUS, nsamples=self.model_cfg.ROI_GRID_POOL.NSAMPLE, mlps=self.model_cfg.ROI_GRID_POOL.MLPS, use_xyz=True, pool_method=self.model_cfg.ROI_GRID_POOL.POOL_METHOD, ) self.class_embed = nn.ModuleList() self.class_embed.append(nn.Linear(model_cfg.Transformer.hidden_dim, 1)) self.bbox_embed = nn.ModuleList() for _ in range(self.num_groups): self.bbox_embed.append(MLP(model_cfg.Transformer.hidden_dim, model_cfg.Transformer.hidden_dim, self.box_coder.code_size * self.num_class, 4)) if self.model_cfg.Transformer.use_grid_pos.enabled: if self.model_cfg.Transformer.use_grid_pos.init_type == 'index': self.grid_index = torch.cat([i.reshape(-1,1)for i in torch.meshgrid(torch.arange(self.grid_size), torch.arange(self.grid_size), torch.arange(self.grid_size))],1).float().cuda() self.grid_pos_embeded = MLP(input_dim = 3, hidden_dim = 256, output_dim = hidden_dim, num_layers = 2) else: self.pos = nn.Parameter(torch.zeros(1, self.num_grid_points, 256)) def init_weights(self, weight_init='xavier'): if weight_init == 'kaiming': init_func = nn.init.kaiming_normal_ elif weight_init == 'xavier': init_func = nn.init.xavier_normal_ elif weight_init == 'normal': init_func = nn.init.normal_ else: raise NotImplementedError for m in self.modules(): if isinstance(m, nn.Conv2d) or isinstance(m, nn.Conv1d): if weight_init == 'normal': init_func(m.weight, mean=0, std=0.001) else: init_func(m.weight) if m.bias is not None: nn.init.constant_(m.bias, 0) nn.init.normal_(self.bbox_embed.layers[-1].weight, mean=0, std=0.001) def get_corner_points_of_roi(self, rois): rois = rois.view(-1, rois.shape[-1]) batch_size_rcnn = rois.shape[0] local_roi_grid_points = self.get_corner_points(rois, batch_size_rcnn) local_roi_grid_points = common_utils.rotate_points_along_z( local_roi_grid_points.clone(), rois[:, 6] ).squeeze(dim=1) global_center = rois[:, 0:3].clone() global_roi_grid_points = local_roi_grid_points + global_center.unsqueeze(dim=1) return global_roi_grid_points, local_roi_grid_points @staticmethod def get_dense_grid_points(rois, batch_size_rcnn, grid_size): if isinstance(grid_size,list): faked_features = rois.new_ones((grid_size[0], grid_size[1], grid_size[2])) grid_size = torch.tensor(grid_size).float().cuda() else: faked_features = rois.new_ones((grid_size, grid_size, grid_size)) dense_idx = faked_features.nonzero() dense_idx = dense_idx.repeat(batch_size_rcnn, 1, 1).float() local_roi_size = rois.view(batch_size_rcnn, -1)[:, 3:6] roi_grid_points = torch.div((dense_idx + 0.5), grid_size) * local_roi_size.unsqueeze(dim=1) - (local_roi_size.unsqueeze(dim=1) / 2) return roi_grid_points @staticmethod def get_corner_points(rois, batch_size_rcnn): faked_features = rois.new_ones((2, 2, 2)) dense_idx = faked_features.nonzero() dense_idx = dense_idx.repeat(batch_size_rcnn, 1, 1).float() local_roi_size = rois.view(batch_size_rcnn, -1)[:, 3:6] roi_grid_points = dense_idx * local_roi_size.unsqueeze(dim=1) \ - (local_roi_size.unsqueeze(dim=1) / 2) return roi_grid_points def get_proxy_points_of_roi(self, rois, grid_size): rois = rois.view(-1, rois.shape[-1]) batch_size_rcnn = rois.shape[0] local_roi_grid_points = self.get_dense_grid_points(rois, batch_size_rcnn, grid_size) local_roi_grid_points = common_utils.rotate_points_along_z(local_roi_grid_points.clone(), rois[:, 6]).squeeze(dim=1) global_center = rois[:, 0:3].clone() global_roi_grid_points = local_roi_grid_points + global_center.unsqueeze(dim=1) return global_roi_grid_points, local_roi_grid_points def roi_grid_pool(self, batch_size, rois, point_coords, point_features,batch_dict=None,batch_cnt=None): """ Args: batch_dict: batch_size: rois: (B, num_rois, 7 + C) point_coords: (num_points, 4) [bs_idx, x, y, z] point_features: (num_points, C) point_cls_scores: (N1 + N2 + N3 + ..., 1) point_part_offset: (N1 + N2 + N3 + ..., 3) Returns: """ num_frames = batch_dict['num_frames'] num_rois = rois.shape[2]*rois.shape[1] global_roi_proxy_points, local_roi_proxy_points = self.get_proxy_points_of_roi( rois.permute(0,2,1,3).contiguous(), grid_size=self.grid_size ) num_points = point_coords.shape[1] num_proxy_points = self.num_proxy_points xyz = point_coords[:, :, 0:3].view(-1,3) if batch_cnt is None: xyz_batch_cnt = torch.tensor([num_points]*rois.shape[2]*batch_size).cuda().int() else: xyz_batch_cnt = torch.tensor(batch_cnt).cuda().int() new_xyz = torch.cat([i[0] for i in global_roi_proxy_points.chunk(rois.shape[2],0)],0) new_xyz_batch_cnt = torch.tensor([self.num_proxy_points]*rois.shape[2]*batch_size).cuda().int() _, pooled_features = self.roi_grid_pool_layer( xyz=xyz.contiguous(), xyz_batch_cnt=xyz_batch_cnt, new_xyz=new_xyz, new_xyz_batch_cnt=new_xyz_batch_cnt, features=point_features.view(-1,point_features.shape[-1]).contiguous(), ) features = pooled_features.view( point_features.shape[0], self.num_proxy_points, pooled_features.shape[-1] ).contiguous() return features,global_roi_proxy_points.view(batch_size*rois.shape[2], num_frames*num_proxy_points,3).contiguous() def spherical_coordinate(self, src, diag_dist): assert (src.shape[-1] == 27) device = src.device indices_x = torch.LongTensor([0,3,6,9,12,15,18,21,24]).to(device) # indices_y = torch.LongTensor([1,4,7,10,13,16,19,22,25]).to(device) # indices_z = torch.LongTensor([2,5,8,11,14,17,20,23,26]).to(device) src_x = torch.index_select(src, -1, indices_x) src_y = torch.index_select(src, -1, indices_y) src_z = torch.index_select(src, -1, indices_z) dis = (src_x ** 2 + src_y ** 2 + src_z ** 2) ** 0.5 phi = torch.atan(src_y / (src_x + 1e-5)) the = torch.acos(src_z / (dis + 1e-5)) dis = dis / (diag_dist + 1e-5) src = torch.cat([dis, phi, the], dim = -1) return src def crop_current_frame_points(self, src, batch_size,trajectory_rois,num_rois,num_sample, batch_dict): for bs_idx in range(batch_size): cur_batch_boxes = trajectory_rois[bs_idx,0,:,:7].view(-1,7) cur_radiis = torch.sqrt((cur_batch_boxes[:,3]/2) ** 2 + (cur_batch_boxes[:,4]/2) ** 2) * 1.1 cur_points = batch_dict['points'][(batch_dict['points'][:, 0] == bs_idx)][:,1:] time_mask = cur_points[:,-1].abs() < 1e-3 cur_points = cur_points[time_mask] dis = torch.norm((cur_points[:,:2].unsqueeze(0) - cur_batch_boxes[:,:2].unsqueeze(1).repeat(1,cur_points.shape[0],1)), dim = 2) point_mask = (dis <= cur_radiis.unsqueeze(-1)) mask = point_mask sampled_idx = torch.topk(mask.float(),128)[1] sampled_idx_buffer = sampled_idx[:, 0:1].repeat(1, 128) roi_idx = torch.arange(num_rois)[:, None].repeat(1, 128) sampled_mask = mask[roi_idx, sampled_idx] sampled_idx_buffer[sampled_mask] = sampled_idx[sampled_mask] src[bs_idx] = cur_points[sampled_idx_buffer][:,:,:5] empty_flag = sampled_mask.sum(-1)==0 src[bs_idx,empty_flag] = 0 return src def trajectories_auxiliary_branch(self,trajectory_rois): time_stamp = torch.ones([trajectory_rois.shape[0],trajectory_rois.shape[1],trajectory_rois.shape[2],1]).cuda() for i in range(time_stamp.shape[1]): time_stamp[:,i,:] = i*0.1 box_seq = torch.cat([trajectory_rois[:,:,:,:7],time_stamp],-1) box_seq[:, :, :,0:3] = box_seq[:, :, :,0:3] - box_seq[:, 0:1, :, 0:3] roi_ry = box_seq[:,:,:,6] % (2 * np.pi) roi_ry_t0 = roi_ry[:,0] roi_ry_t0 = roi_ry_t0.repeat(1,box_seq.shape[1]) # transfer LiDAR coords to local coords box_seq = common_utils.rotate_points_along_z( points=box_seq.view(-1, 1, box_seq.shape[-1]), angle=-roi_ry_t0.view(-1) ).view(box_seq.shape[0],box_seq.shape[1], -1, box_seq.shape[-1]) box_seq[:,:,:,6] = 0 batch_rcnn = box_seq.shape[0]*box_seq.shape[2] box_reg, box_feat, _ = self.seqboxembed(box_seq.permute(0,2,3,1).contiguous().view(batch_rcnn,box_seq.shape[-1],box_seq.shape[1])) return box_reg, box_feat def get_proposal_aware_motion_feature(self,proxy_point,batch_size,trajectory_rois,num_rois,batch_dict): time_stamp = torch.ones([proxy_point.shape[0],proxy_point.shape[1],1]).cuda() padding_zero = torch.zeros([proxy_point.shape[0],proxy_point.shape[1],2]).cuda() proxy_point_padding = torch.cat([padding_zero,time_stamp],-1) num_time_coding = trajectory_rois.shape[1] for i in range(num_time_coding): proxy_point_padding[:,i*self.num_proxy_points:(i+1)*self.num_proxy_points,-1] = i*0.1 ######### use T0 Norm ######## corner_points, _ = self.get_corner_points_of_roi(trajectory_rois[:,0,:,:].contiguous()) corner_points = corner_points.view(batch_size, num_rois, -1, corner_points.shape[-1]) corner_points = corner_points.view(batch_size * num_rois, -1) corner_add_center_points = torch.cat([corner_points, trajectory_rois[:,0,:,:].reshape(batch_size * num_rois, -1)[:,:3]], dim = -1) pos_fea = proxy_point[:,:,:3].repeat(1,1,9) - corner_add_center_points.unsqueeze(1) lwh = trajectory_rois[:,0,:,:].reshape(batch_size * num_rois, -1)[:,3:6].unsqueeze(1).repeat(1,proxy_point.shape[1],1) diag_dist = (lwh[:,:,0]**2 + lwh[:,:,1]**2 + lwh[:,:,2]**2) ** 0.5 pos_fea = self.spherical_coordinate(pos_fea, diag_dist = diag_dist.unsqueeze(-1)) ######### use T0 Norm ######## proxy_point_padding = torch.cat([pos_fea,proxy_point_padding],-1) proxy_point_motion_feat = self.up_dimension_motion(proxy_point_padding) return proxy_point_motion_feat def get_proposal_aware_geometry_feature(self,src, batch_size,trajectory_rois,num_rois,batch_dict): i = 0 # only current frame corner_points, _ = self.get_corner_points_of_roi(trajectory_rois[:,i,:,:].contiguous()) corner_points = corner_points.view(batch_size, num_rois, -1, corner_points.shape[-1]) corner_points = corner_points.view(batch_size * num_rois, -1) trajectory_roi_center = trajectory_rois[:,i,:,:].contiguous().reshape(batch_size * num_rois, -1)[:,:3] corner_add_center_points = torch.cat([corner_points, trajectory_roi_center], dim = -1) proposal_aware_feat = src[:,i*self.num_lidar_points:(i+1)*self.num_lidar_points,:3].repeat(1,1,9) - \ corner_add_center_points.unsqueeze(1).repeat(1,self.num_lidar_points,1) lwh = trajectory_rois[:,i,:,:].reshape(batch_size * num_rois, -1)[:,3:6].unsqueeze(1).repeat(1,proposal_aware_feat.shape[1],1) diag_dist = (lwh[:,:,0]**2 + lwh[:,:,1]**2 + lwh[:,:,2]**2) ** 0.5 proposal_aware_feat = self.spherical_coordinate(proposal_aware_feat, diag_dist = diag_dist.unsqueeze(-1)) proposal_aware_feat = torch.cat([proposal_aware_feat, src[:,:,3:]], dim = -1) src_gemoetry = self.up_dimension_geometry(proposal_aware_feat) proxy_point_geometry, proxy_points = self.roi_grid_pool(batch_size,trajectory_rois,src,src_gemoetry,batch_dict,batch_cnt=None) return proxy_point_geometry,proxy_points @staticmethod def reorder_rois_for_refining(pred_bboxes): num_max_rois = max([len(bbox) for bbox in pred_bboxes]) num_max_rois = max(1, num_max_rois) # at least one faked rois to avoid error ordered_bboxes = torch.zeros([len(pred_bboxes),num_max_rois,pred_bboxes[0].shape[-1]]).cuda() for bs_idx in range(ordered_bboxes.shape[0]): ordered_bboxes[bs_idx,:len(pred_bboxes[bs_idx])] = pred_bboxes[bs_idx] return ordered_bboxes def transform_prebox_to_current_vel(self,pred_boxes3d,pose_pre,pose_cur): expand_bboxes = np.concatenate([pred_boxes3d[:,:3], np.ones((pred_boxes3d.shape[0], 1))], axis=-1) expand_vels = np.concatenate([pred_boxes3d[:,7:9], np.zeros((pred_boxes3d.shape[0], 1))], axis=-1) bboxes_global = np.dot(expand_bboxes, pose_pre.T)[:, :3] vels_global = np.dot(expand_vels, pose_pre[:3,:3].T) moved_bboxes_global = copy.deepcopy(bboxes_global) moved_bboxes_global[:,:2] = moved_bboxes_global[:,:2] - 0.1*vels_global[:,:2] expand_bboxes_global = np.concatenate([bboxes_global[:,:3],np.ones((bboxes_global.shape[0], 1))], axis=-1) expand_moved_bboxes_global = np.concatenate([moved_bboxes_global[:,:3],np.ones((bboxes_global.shape[0], 1))], axis=-1) bboxes_pre2cur = np.dot(expand_bboxes_global, np.linalg.inv(pose_cur.T))[:, :3] moved_bboxes_pre2cur = np.dot(expand_moved_bboxes_global, np.linalg.inv(pose_cur.T))[:, :3] vels_pre2cur = np.dot(vels_global, np.linalg.inv(pose_cur[:3,:3].T))[:,:2] bboxes_pre2cur = np.concatenate([bboxes_pre2cur, pred_boxes3d[:,3:7],vels_pre2cur],axis=-1) bboxes_pre2cur[:,6] = bboxes_pre2cur[..., 6] + np.arctan2(pose_pre[1, 0], pose_pre[0,0]) bboxes_pre2cur[:,6] = bboxes_pre2cur[..., 6] - np.arctan2(pose_cur[1, 0], pose_cur[0,0]) bboxes_pre2cur[:,7:9] = moved_bboxes_pre2cur[:,:2] - bboxes_pre2cur[:,:2] return bboxes_pre2cur[None,:,:] def generate_trajectory(self,cur_batch_boxes,proposals_list,batch_dict): trajectory_rois = cur_batch_boxes[:,None,:,:].repeat(1,batch_dict['rois'].shape[-2],1,1) trajectory_rois[:,0,:,:]= cur_batch_boxes valid_length = torch.zeros([batch_dict['batch_size'],batch_dict['rois'].shape[-2],trajectory_rois.shape[2]]) valid_length[:,0] = 1 num_frames = batch_dict['rois'].shape[-2] matching_table = (trajectory_rois.new_ones([trajectory_rois.shape[1],trajectory_rois.shape[2]]) * -1).long() for i in range(1,num_frames): frame = torch.zeros_like(cur_batch_boxes) frame[:,:,0:2] = trajectory_rois[:,i-1,:,0:2] + trajectory_rois[:,i-1,:,7:9] frame[:,:,2:] = trajectory_rois[:,i-1,:,2:] for bs_idx in range( batch_dict['batch_size']): iou3d = iou3d_nms_utils.boxes_iou3d_gpu(frame[bs_idx,:,:7], proposals_list[bs_idx,i,:,:7]) max_overlaps, traj_assignment = torch.max(iou3d, dim=1) fg_inds = ((max_overlaps >= 0.5)).nonzero().view(-1) valid_length[bs_idx,i,fg_inds] = 1 matching_table[i,fg_inds] = traj_assignment[fg_inds] trajectory_rois[bs_idx,i,fg_inds,:] = proposals_list[bs_idx,i,traj_assignment[fg_inds]] batch_dict['valid_length'] = valid_length return trajectory_rois,valid_length, matching_table def forward(self, batch_dict): """ :param input_data: input dict :return: """ if 'memory_bank' in batch_dict.keys(): rois_list = [] memory_list = copy.deepcopy(batch_dict['memory_bank']) for idx in range(len(memory_list['rois'])): rois = torch.cat([batch_dict['memory_bank']['rois'][idx][0], batch_dict['memory_bank']['roi_scores'][idx][0], batch_dict['memory_bank']['roi_labels'][idx][0]],-1) rois_list.append(rois) batch_rois = self.reorder_rois_for_refining(rois_list) batch_dict['roi_scores'] = batch_rois[None,:,:,9] batch_dict['roi_labels'] = batch_rois[None,:,:,10] proposals_list = [] for i in range(self.model_cfg.Transformer.num_frames): pose_pre = batch_dict['poses'][0,i*4:(i+1)*4,:] pred2cur = self.transform_prebox_to_current_vel(batch_rois[i,:,:9].cpu().numpy(),pose_pre=pose_pre.cpu().numpy(), pose_cur=batch_dict['poses'][0,:4,:].cpu().numpy()) proposals_list.append(torch.from_numpy(pred2cur).cuda().float()) batch_rois = torch.cat(proposals_list,0) batch_dict['proposals_list'] = batch_rois[None,:,:,:9] batch_dict['rois'] = batch_rois.unsqueeze(0).permute(0,2,1,3) num_rois = batch_dict['rois'].shape[1] batch_dict['num_frames'] = batch_dict['rois'].shape[2] roi_labels_list = copy.deepcopy(batch_dict['roi_labels']) batch_dict['roi_scores'] = batch_dict['roi_scores'].permute(0,2,1) batch_dict['roi_labels'] = batch_dict['roi_labels'][:,0,:].long() proposals_list = batch_dict['proposals_list'] batch_size = batch_dict['batch_size'] cur_batch_boxes = copy.deepcopy(batch_dict['rois'].detach())[:,:,0] batch_dict['cur_frame_idx'] = 0 else: batch_dict['rois'] = batch_dict['proposals_list'].permute(0,2,1,3) assert batch_dict['rois'].shape[0] ==1 num_rois = batch_dict['rois'].shape[1] batch_dict['num_frames'] = batch_dict['rois'].shape[2] roi_labels_list = copy.deepcopy(batch_dict['roi_labels']) batch_dict['roi_scores'] = batch_dict['roi_scores'].permute(0,2,1) batch_dict['roi_labels'] = batch_dict['roi_labels'][:,0,:].long() proposals_list = batch_dict['proposals_list'] batch_size = batch_dict['batch_size'] cur_batch_boxes = copy.deepcopy(batch_dict['rois'].detach())[:,:,0] batch_dict['cur_frame_idx'] = 0 trajectory_rois,effective_length,matching_table = self.generate_trajectory(cur_batch_boxes,proposals_list,batch_dict) batch_dict['has_class_labels'] = True batch_dict['trajectory_rois'] = trajectory_rois rois = batch_dict['rois'] num_rois = batch_dict['rois'].shape[1] if self.model_cfg.get('USE_TRAJ_EMPTY_MASK',None): empty_mask = batch_dict['rois'][:,:,0,:6].sum(-1)==0 batch_dict['valid_traj_mask'] = ~empty_mask num_sample = self.num_lidar_points src = rois.new_zeros(batch_size, num_rois, num_sample, 5) src = self.crop_current_frame_points(src, batch_size, trajectory_rois, num_rois, num_sample, batch_dict) src = src.view(batch_size * num_rois, -1, src.shape[-1]) src_geometry_feature,proxy_points = self.get_proposal_aware_geometry_feature(src,batch_size,trajectory_rois,num_rois,batch_dict) src_motion_feature = self.get_proposal_aware_motion_feature(proxy_points,batch_size,trajectory_rois,num_rois,batch_dict) if batch_dict['sample_idx'][0] >=1: src_repeat = src_geometry_feature[:,None,:self.num_proxy_points,:].repeat([1,trajectory_rois.shape[1],1,1]) src_before = src_repeat[:,1:,:,:].clone() #[bs,traj,num_roi,C] valid_length = batch_dict['num_frames'] -1 if batch_dict['sample_idx'][0] > batch_dict['num_frames'] -1 \ else int(batch_dict['sample_idx'][0].item()) num_max_rois = max(trajectory_rois.shape[2], *[i.shape[0] for i in batch_dict['memory_bank']['feature_bank']]) feature_bank = self.reorder_memory(batch_dict['memory_bank']['feature_bank'][:valid_length],num_max_rois) effective_length = effective_length[0,1:1+valid_length].bool() #rm dim of bs for i in range(valid_length): src_before[:,i][effective_length[i]] = feature_bank[i,matching_table[1+i][effective_length[i]]] src_geometry_feature = torch.cat([src_repeat[:,:1],src_before],1).view(src_geometry_feature.shape[0],-1, src_geometry_feature.shape[-1]) else: src_geometry_feature = src_geometry_feature.repeat([1,trajectory_rois.shape[1],1]) batch_dict['geometory_feature_memory'] = src_geometry_feature[:,:self.num_proxy_points] src = src_geometry_feature + src_motion_feature if self.model_cfg.get('USE_TRAJ_EMPTY_MASK',None): src[empty_mask.view(-1)] = 0 if self.model_cfg.Transformer.use_grid_pos.init_type == 'index': pos = self.grid_pos_embeded(self.grid_index.cuda())[None,:,:] pos = torch.cat([torch.zeros(1,1,self.hidden_dim).cuda(),pos],1) else: pos=None hs, tokens = self.transformer(src,pos=pos) point_cls_list = [] for i in range(self.num_enc_layer): point_cls_list.append(self.class_embed[0](tokens[i][0])) point_cls = torch.cat(point_cls_list,0) hs = hs.permute(1,0,2).reshape(hs.shape[1],-1) _, feat_box = self.trajectories_auxiliary_branch(trajectory_rois) joint_reg = self.jointembed(torch.cat([hs,feat_box],-1)) rcnn_cls = point_cls rcnn_reg = joint_reg if not self.training: batch_dict['rois'] = batch_dict['rois'][:,:,0].contiguous() rcnn_cls = rcnn_cls[-rcnn_cls.shape[0]//self.num_enc_layer:] batch_cls_preds, batch_box_preds = self.generate_predicted_boxes( batch_size=batch_dict['batch_size'], rois=batch_dict['rois'], cls_preds=rcnn_cls, box_preds=rcnn_reg ) batch_dict['batch_box_preds'] = batch_box_preds batch_dict['cls_preds_normalized'] = False if self.avg_stage1_score: stage1_score = batch_dict['roi_scores'][:,:,:1] batch_cls_preds = F.sigmoid(batch_cls_preds) if self.model_cfg.get('IOU_WEIGHT', None): batch_box_preds_list = [] roi_labels_list = [] batch_cls_preds_list = [] for bs_idx in range(batch_size): car_mask = batch_dict['roi_labels'][bs_idx] ==1 batch_cls_preds_car = batch_cls_preds[bs_idx].pow(self.model_cfg.IOU_WEIGHT[0])* \ stage1_score[bs_idx].pow(1-self.model_cfg.IOU_WEIGHT[0]) batch_cls_preds_car = batch_cls_preds_car[car_mask][None] batch_cls_preds_pedcyc = batch_cls_preds[bs_idx].pow(self.model_cfg.IOU_WEIGHT[1])* \ stage1_score[bs_idx].pow(1-self.model_cfg.IOU_WEIGHT[1]) batch_cls_preds_pedcyc = batch_cls_preds_pedcyc[~car_mask][None] cls_preds = torch.cat([batch_cls_preds_car,batch_cls_preds_pedcyc],1) box_preds = torch.cat([batch_dict['batch_box_preds'][bs_idx][car_mask], batch_dict['batch_box_preds'][bs_idx][~car_mask]],0)[None] roi_labels = torch.cat([batch_dict['roi_labels'][bs_idx][car_mask], batch_dict['roi_labels'][bs_idx][~car_mask]],0)[None] batch_box_preds_list.append(box_preds) roi_labels_list.append(roi_labels) batch_cls_preds_list.append(cls_preds) batch_dict['batch_box_preds'] = torch.cat(batch_box_preds_list,0) batch_dict['roi_labels'] = torch.cat(roi_labels_list,0) batch_cls_preds = torch.cat(batch_cls_preds_list,0) else: batch_cls_preds = torch.sqrt(batch_cls_preds*stage1_score) batch_dict['cls_preds_normalized'] = True batch_dict['batch_cls_preds'] = batch_cls_preds return batch_dict def reorder_memory(self, memory,num_max_rois): ordered_memory = memory[0].new_zeros([len(memory),num_max_rois,memory[0].shape[1],memory[0].shape[2]]) for bs_idx in range(len(memory)): ordered_memory[bs_idx,:len(memory[bs_idx])] = memory[bs_idx] return ordered_memory def generate_predicted_boxes(self, batch_size, rois, cls_preds=None, box_preds=None): """ Args: batch_size: rois: (B, N, 7) cls_preds: (BN, num_class) box_preds: (BN, code_size) Returns: """ code_size = self.box_coder.code_size if cls_preds is not None: batch_cls_preds = cls_preds.view(batch_size, -1, cls_preds.shape[-1]) else: batch_cls_preds = None batch_box_preds = box_preds.view(batch_size, -1, code_size) roi_ry = rois[:, :, 6].view(-1) roi_xyz = rois[:, :, 0:3].view(-1, 3) local_rois = rois.clone().detach() local_rois[:, :, 0:3] = 0 batch_box_preds = self.box_coder.decode_torch(batch_box_preds, local_rois).view(-1, code_size) batch_box_preds = common_utils.rotate_points_along_z( batch_box_preds.unsqueeze(dim=1), roi_ry ).squeeze(dim=1) batch_box_preds[:, 0:3] += roi_xyz batch_box_preds = batch_box_preds.view(batch_size, -1, code_size) batch_box_preds = torch.cat([batch_box_preds,rois[:,:,7:]],-1) return batch_cls_preds, batch_box_preds