search

sshaoshuai / PointRCNN

PointRCNN: 3D Object Proposal Generation and Detection from Point Cloud, CVPR 2019.
MIT License
1.71k stars 426 forks source link

eval_rcnn.py RuntimeError: shape mismatch #106

Open honglongcai opened 5 years ago

honglongcai commented 5 years ago

2019-09-13 11:27:40,282 INFO **Start logging** 2019-09-13 11:27:40,283 INFO cfg_file cfgs/default.yaml 2019-09-13 11:27:40,283 INFO eval_mode rcnn 2019-09-13 11:27:40,283 INFO eval_all False 2019-09-13 11:27:40,283 INFO test False 2019-09-13 11:27:40,283 INFO ckpt ../models/PointRCNN.pth 2019-09-13 11:27:40,283 INFO rpn_ckpt None 2019-09-13 11:27:40,283 INFO rcnn_ckpt None 2019-09-13 11:27:40,283 INFO batch_size 1 2019-09-13 11:27:40,283 INFO workers 4 2019-09-13 11:27:40,283 INFO extra_tag default 2019-09-13 11:27:40,283 INFO output_dir None 2019-09-13 11:27:40,283 INFO ckpt_dir None 2019-09-13 11:27:40,283 INFO save_result False 2019-09-13 11:27:40,283 INFO save_rpn_feature False 2019-09-13 11:27:40,283 INFO random_select True 2019-09-13 11:27:40,283 INFO start_epoch 0 2019-09-13 11:27:40,284 INFO rcnn_eval_roi_dir None 2019-09-13 11:27:40,284 INFO rcnn_eval_feature_dir None 2019-09-13 11:27:40,284 INFO set_cfgs ['RPN.LOC_XZ_FINE', 'False'] 2019-09-13 11:27:40,284 INFO cfg.TAG: default 2019-09-13 11:27:40,284 INFO cfg.CLASSES: Car 2019-09-13 11:27:40,284 INFO cfg.INCLUDE_SIMILAR_TYPE: True 2019-09-13 11:27:40,284 INFO cfg.AUG_DATA: True 2019-09-13 11:27:40,284 INFO cfg.AUG_METHOD_LIST: ['rotation', 'scaling', 'flip'] 2019-09-13 11:27:40,284 INFO cfg.AUG_METHOD_PROB: [1.0, 1.0, 0.5] 2019-09-13 11:27:40,284 INFO cfg.AUG_ROT_RANGE: 18 2019-09-13 11:27:40,284 INFO cfg.GT_AUG_ENABLED: True 2019-09-13 11:27:40,284 INFO cfg.GT_EXTRA_NUM: 15 2019-09-13 11:27:40,284 INFO cfg.GT_AUG_RAND_NUM: True 2019-09-13 11:27:40,284 INFO cfg.GT_AUG_APPLY_PROB: 1.0 2019-09-13 11:27:40,284 INFO cfg.GT_AUG_HARD_RATIO: 0.6 2019-09-13 11:27:40,284 INFO cfg.PC_REDUCE_BY_RANGE: True 2019-09-13 11:27:40,285 INFO cfg.PC_AREA_SCOPE: [[-40. 40. ] [ -1. 3. ] [ 0. 70.4]] 2019-09-13 11:27:40,285 INFO cfg.CLS_MEAN_SIZE: [[1.5256319 1.6285675 3.8831165]] 2019-09-13 11:27:40,285 INFO
cfg.RPN = edict() 2019-09-13 11:27:40,285 INFO cfg.RPN.ENABLED: True 2019-09-13 11:27:40,285 INFO cfg.RPN.FIXED: True 2019-09-13 11:27:40,285 INFO cfg.RPN.USE_INTENSITY: False 2019-09-13 11:27:40,285 INFO cfg.RPN.LOC_XZ_FINE: False 2019-09-13 11:27:40,285 INFO cfg.RPN.LOC_SCOPE: 3.0 2019-09-13 11:27:40,285 INFO cfg.RPN.LOC_BIN_SIZE: 0.5 2019-09-13 11:27:40,285 INFO cfg.RPN.NUM_HEAD_BIN: 12 2019-09-13 11:27:40,285 INFO cfg.RPN.BACKBONE: pointnet2_msg 2019-09-13 11:27:40,285 INFO cfg.RPN.USE_BN: True 2019-09-13 11:27:40,285 INFO cfg.RPN.NUM_POINTS: 16384 2019-09-13 11:27:40,285 INFO
cfg.RPN.SA_CONFIG = edict() 2019-09-13 11:27:40,285 INFO cfg.RPN.SA_CONFIG.NPOINTS: [4096, 1024, 256, 64] 2019-09-13 11:27:40,285 INFO cfg.RPN.SA_CONFIG.RADIUS: [[0.1, 0.5], [0.5, 1.0], [1.0, 2.0], [2.0, 4.0]] 2019-09-13 11:27:40,286 INFO cfg.RPN.SA_CONFIG.NSAMPLE: [[16, 32], [16, 32], [16, 32], [16, 32]] 2019-09-13 11:27:40,286 INFO cfg.RPN.SA_CONFIG.MLPS: [[[16, 16, 32], [32, 32, 64]], [[64, 64, 128], [64, 96, 128]], [[128, 196, 256], [128, 196, 256]], [[256, 256, 512], [256, 384, 512]]] 2019-09-13 11:27:40,286 INFO cfg.RPN.FP_MLPS: [[128, 128], [256, 256], [512, 512], [512, 512]] 2019-09-13 11:27:40,286 INFO cfg.RPN.CLS_FC: [128] 2019-09-13 11:27:40,286 INFO cfg.RPN.REG_FC: [128] 2019-09-13 11:27:40,286 INFO cfg.RPN.DP_RATIO: 0.5 2019-09-13 11:27:40,286 INFO cfg.RPN.LOSS_CLS: SigmoidFocalLoss 2019-09-13 11:27:40,286 INFO cfg.RPN.FG_WEIGHT: 15 2019-09-13 11:27:40,286 INFO cfg.RPN.FOCAL_ALPHA: [0.25, 0.75] 2019-09-13 11:27:40,286 INFO cfg.RPN.FOCAL_GAMMA: 2.0 2019-09-13 11:27:40,286 INFO cfg.RPN.REG_LOSS_WEIGHT: [1.0, 1.0, 1.0, 1.0] 2019-09-13 11:27:40,286 INFO cfg.RPN.LOSS_WEIGHT: [1.0, 1.0] 2019-09-13 11:27:40,286 INFO cfg.RPN.NMS_TYPE: normal 2019-09-13 11:27:40,286 INFO cfg.RPN.SCORE_THRESH: 0.3 2019-09-13 11:27:40,286 INFO
cfg.RCNN = edict() 2019-09-13 11:27:40,286 INFO cfg.RCNN.ENABLED: True 2019-09-13 11:27:40,286 INFO cfg.RCNN.USE_RPN_FEATURES: True 2019-09-13 11:27:40,286 INFO cfg.RCNN.USE_MASK: True 2019-09-13 11:27:40,286 INFO cfg.RCNN.MASK_TYPE: seg 2019-09-13 11:27:40,286 INFO cfg.RCNN.USE_INTENSITY: False 2019-09-13 11:27:40,287 INFO cfg.RCNN.USE_DEPTH: True 2019-09-13 11:27:40,287 INFO cfg.RCNN.USE_SEG_SCORE: False 2019-09-13 11:27:40,287 INFO cfg.RCNN.ROI_SAMPLE_JIT: True 2019-09-13 11:27:40,287 INFO cfg.RCNN.ROI_FG_AUG_TIMES: 10 2019-09-13 11:27:40,287 INFO cfg.RCNN.REG_AUG_METHOD: multiple 2019-09-13 11:27:40,287 INFO cfg.RCNN.POOL_EXTRA_WIDTH: 1.0 2019-09-13 11:27:40,287 INFO cfg.RCNN.LOC_SCOPE: 1.5 2019-09-13 11:27:40,287 INFO cfg.RCNN.LOC_BIN_SIZE: 0.5 2019-09-13 11:27:40,287 INFO cfg.RCNN.NUM_HEAD_BIN: 9 2019-09-13 11:27:40,287 INFO cfg.RCNN.LOC_Y_BY_BIN: False 2019-09-13 11:27:40,287 INFO cfg.RCNN.LOC_Y_SCOPE: 0.5 2019-09-13 11:27:40,287 INFO cfg.RCNN.LOC_Y_BIN_SIZE: 0.25 2019-09-13 11:27:40,287 INFO cfg.RCNN.SIZE_RES_ON_ROI: False 2019-09-13 11:27:40,287 INFO cfg.RCNN.USE_BN: False 2019-09-13 11:27:40,287 INFO cfg.RCNN.DP_RATIO: 0.0 2019-09-13 11:27:40,287 INFO cfg.RCNN.BACKBONE: pointnet 2019-09-13 11:27:40,287 INFO cfg.RCNN.XYZ_UP_LAYER: [128, 128] 2019-09-13 11:27:40,287 INFO cfg.RCNN.NUM_POINTS: 512 2019-09-13 11:27:40,287 INFO
cfg.RCNN.SA_CONFIG = edict() 2019-09-13 11:27:40,287 INFO cfg.RCNN.SA_CONFIG.NPOINTS: [128, 32, -1] 2019-09-13 11:27:40,288 INFO cfg.RCNN.SA_CONFIG.RADIUS: [0.2, 0.4, 100] 2019-09-13 11:27:40,288 INFO cfg.RCNN.SA_CONFIG.NSAMPLE: [64, 64, 64] 2019-09-13 11:27:40,288 INFO cfg.RCNN.SA_CONFIG.MLPS: [[128, 128, 128], [128, 128, 256], [256, 256, 512]] 2019-09-13 11:27:40,288 INFO cfg.RCNN.CLS_FC: [256, 256] 2019-09-13 11:27:40,288 INFO cfg.RCNN.REG_FC: [256, 256] 2019-09-13 11:27:40,288 INFO cfg.RCNN.LOSS_CLS: BinaryCrossEntropy 2019-09-13 11:27:40,288 INFO cfg.RCNN.FOCAL_ALPHA: [0.25, 0.75] 2019-09-13 11:27:40,288 INFO cfg.RCNN.FOCAL_GAMMA: 2.0 2019-09-13 11:27:40,288 INFO cfg.RCNN.CLS_WEIGHT: [1. 1. 1.] 2019-09-13 11:27:40,288 INFO cfg.RCNN.CLS_FG_THRESH: 0.6 2019-09-13 11:27:40,288 INFO cfg.RCNN.CLS_BG_THRESH: 0.45 2019-09-13 11:27:40,288 INFO cfg.RCNN.CLS_BG_THRESH_LO: 0.05 2019-09-13 11:27:40,288 INFO cfg.RCNN.REG_FG_THRESH: 0.55 2019-09-13 11:27:40,288 INFO cfg.RCNN.FG_RATIO: 0.5 2019-09-13 11:27:40,288 INFO cfg.RCNN.ROI_PER_IMAGE: 64 2019-09-13 11:27:40,288 INFO cfg.RCNN.HARD_BG_RATIO: 0.8 2019-09-13 11:27:40,288 INFO cfg.RCNN.SCORE_THRESH: 0.3 2019-09-13 11:27:40,289 INFO cfg.RCNN.NMS_THRESH: 0.1 2019-09-13 11:27:40,289 INFO
cfg.TRAIN = edict() 2019-09-13 11:27:40,289 INFO cfg.TRAIN.SPLIT: train 2019-09-13 11:27:40,289 INFO cfg.TRAIN.VAL_SPLIT: smallval 2019-09-13 11:27:40,289 INFO cfg.TRAIN.LR: 0.002 2019-09-13 11:27:40,289 INFO cfg.TRAIN.LR_CLIP: 1e-05 2019-09-13 11:27:40,289 INFO cfg.TRAIN.LR_DECAY: 0.5 2019-09-13 11:27:40,289 INFO cfg.TRAIN.DECAY_STEP_LIST: [100, 150, 180, 200] 2019-09-13 11:27:40,289 INFO cfg.TRAIN.LR_WARMUP: True 2019-09-13 11:27:40,289 INFO cfg.TRAIN.WARMUP_MIN: 0.0002 2019-09-13 11:27:40,289 INFO cfg.TRAIN.WARMUP_EPOCH: 1 2019-09-13 11:27:40,289 INFO cfg.TRAIN.BN_MOMENTUM: 0.1 2019-09-13 11:27:40,289 INFO cfg.TRAIN.BN_DECAY: 0.5 2019-09-13 11:27:40,289 INFO cfg.TRAIN.BNM_CLIP: 0.01 2019-09-13 11:27:40,289 INFO cfg.TRAIN.BN_DECAY_STEP_LIST: [1000] 2019-09-13 11:27:40,289 INFO cfg.TRAIN.OPTIMIZER: adam_onecycle 2019-09-13 11:27:40,289 INFO cfg.TRAIN.WEIGHT_DECAY: 0.001 2019-09-13 11:27:40,289 INFO cfg.TRAIN.MOMENTUM: 0.9 2019-09-13 11:27:40,289 INFO cfg.TRAIN.MOMS: [0.95, 0.85] 2019-09-13 11:27:40,289 INFO cfg.TRAIN.DIV_FACTOR: 10.0 2019-09-13 11:27:40,290 INFO cfg.TRAIN.PCT_START: 0.4 2019-09-13 11:27:40,290 INFO cfg.TRAIN.GRAD_NORM_CLIP: 1.0 2019-09-13 11:27:40,290 INFO cfg.TRAIN.RPN_PRE_NMS_TOP_N: 9000 2019-09-13 11:27:40,290 INFO cfg.TRAIN.RPN_POST_NMS_TOP_N: 512 2019-09-13 11:27:40,290 INFO cfg.TRAIN.RPN_NMS_THRESH: 0.85 2019-09-13 11:27:40,290 INFO cfg.TRAIN.RPN_DISTANCE_BASED_PROPOSE: True 2019-09-13 11:27:40,290 INFO
cfg.TEST = edict() 2019-09-13 11:27:40,290 INFO cfg.TEST.SPLIT: val 2019-09-13 11:27:40,290 INFO cfg.TEST.RPN_PRE_NMS_TOP_N: 9000 2019-09-13 11:27:40,290 INFO cfg.TEST.RPN_POST_NMS_TOP_N: 100 2019-09-13 11:27:40,290 INFO cfg.TEST.RPN_NMS_THRESH: 0.8 2019-09-13 11:27:40,290 INFO cfg.TEST.RPN_DISTANCE_BASED_PROPOSE: True 2019-09-13 11:27:40,292 INFO Load testing samples from ../data/KITTI/object/training 2019-09-13 11:27:40,292 INFO Done: total test samples 3769 2019-09-13 11:27:43,336 INFO ==> Loading from checkpoint '../models/PointRCNN.pth' 2019-09-13 11:27:43,413 INFO ==> Done 2019-09-13 11:27:43,414 INFO ---- EPOCH no_number JOINT EVALUATION ---- 2019-09-13 11:27:43,414 INFO ==> Output file: ../output/rcnn/default/eval/epoch_no_number/val eval: 48%|▍| 1827/3769 [06:46<25:41, 1.26it/s, mode=EVAL, recall=1441151880800eval: 49%|▍| 1828/3769 [06:46<4:45:46, 8.83s/it, mode=EVAL, recall=144115188080092746/6871]Traceback (most recent call last): File "eval_rcnn.py", line 902, in eval_single_ckpt(root_result_dir) File "eval_rcnn.py", line 765, in eval_single_ckpt eval_one_epoch(model, test_loader, epoch_id, root_result_dir, logger) File "eval_rcnn.py", line 692, in eval_one_epoch ret_dict = eval_one_epoch_joint(model, dataloader, epoch_id, result_dir, logger) File "eval_rcnn.py", line 495, in eval_one_epoch_joint ret_dict = model(input_data) File "/home/honglongcai/anaconda3/envs/pytorch1/lib/python3.7/site-packages/torch/nn/modules/module.py", line 489, in call result = self.forward(*input, *kwargs) File "/home/honglongcai/Github/PointRCNN/tools/../lib/net/point_rcnn.py", line 48, in forward rois, roi_scores_raw = self.rpn.proposal_layer(rpn_scores_raw, rpn_reg, backbone_xyz) # (B, M, 7) File "/home/honglongcai/anaconda3/envs/pytorch1/lib/python3.7/site-packages/torch/nn/modules/module.py", line 489, in call result = self.forward(input, *kwargs) File "/home/honglongcai/Github/PointRCNN/tools/../lib/rpn/proposal_layer.py", line 30, in forward get_ry_fine=False) # (N, 7) File "/home/honglongcai/Github/PointRCNN/tools/../lib/utils/bbox_transform.py", line 102, in decode_bbox_target ry[ry > np.pi] -= 2 np.pi RuntimeError: shape mismatch: value tensor of shape [4231196] cannot be broadcast to indexing result of shape [334] eval: 49%|▍| 1828/3769 [06:54<07:20, 4.41it/s, mode=EVAL, recall=144115188080092746/6871]

onecodeonehair commented 1 year ago

did you solved it,i have the same question