Hugging Face's logo

6DammK9
/
cagroup3d-win10-scannet

TensorBoard
English
pcdet
Model card Files Metrics Community
6DammK9 commited on
Commit
467e640
·
1 Parent(s): 65b9122

Upload 6 files

Browse files
Files changed (6) hide show
  1. CAGroup3D.yaml +103 -0
  2. ckpt/checkpoint_epoch_1.pth +3 -0
  3. eval/eval_with_train/eval_list_val.txt +0 -0
  4. eval/eval_with_train/tensorboard_val/events.out.tfevents.1680187773.DESKTOP-3FL13RB +3 -0
  5. log_train_20230326-130440.txt +912 -0
  6. tensorboard/events.out.tfevents.1679807081.DESKTOP-3FL13RB +3 -0
CAGroup3D.yaml ADDED
@@ -0,0 +1,103 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ CLASS_NAMES: [ 'cabinet', 'bed', 'chair', 'sofa', 'table', 'door', 'window',
2
+ 'bookshelf', 'picture', 'counter', 'desk', 'curtain',
3
+ 'refrigerator', 'showercurtrain', 'toilet', 'sink', 'bathtub',
4
+ 'garbagebin']
5
+
6
+ DATA_CONFIG:
7
+ _BASE_CONFIG_: cfgs/dataset_configs/scannet_dataset.yaml
8
+
9
+ VOXEL_SIZE: &VOXEL_SIZE 0.02
10
+ N_CLASSES: &N_CLASSES 18
11
+ SEMANTIC_THR: &SEMANTIC_THR 0.15
12
+
13
+ MODEL:
14
+ NAME: CAGroup3D
15
+ VOXEL_SIZE: *VOXEL_SIZE
16
+ SEMANTIC_MIN_THR: 0.05
17
+ SEMANTIC_ITER_VALUE: 0.02
18
+ SEMANTIC_THR: *SEMANTIC_THR
19
+ BACKBONE_3D:
20
+ NAME: BiResNet
21
+ IN_CHANNELS: 3
22
+ OUT_CHANNELS: 64
23
+
24
+ DENSE_HEAD:
25
+ NAME: CAGroup3DHead
26
+ IN_CHANNELS: [64, 128, 256, 512]
27
+ OUT_CHANNELS: 64
28
+ SEMANTIC_THR: *SEMANTIC_THR
29
+ VOXEL_SIZE: *VOXEL_SIZE
30
+ N_CLASSES: *N_CLASSES
31
+ N_REG_OUTS: 6
32
+ CLS_KERNEL: 9
33
+ WITH_YAW: False
34
+ USE_SEM_SCORE: False # if feed sem scores to the second-stage, default: False
35
+ EXPAND_RATIO: 3
36
+ ASSIGNER:
37
+ NAME: CAGroup3DAssigner
38
+ LIMIT: 27
39
+ TOPK: 18
40
+ N_SCALES: 4
41
+ LOSS_OFFSET:
42
+ NAME: SmoothL1Loss
43
+ BETA: 0.04
44
+ REDUCTION: sum
45
+ LOSS_WEIGHT: 1.0
46
+ LOSS_BBOX:
47
+ NAME: IoU3DLoss
48
+ WITH_YAW: False
49
+ LOSS_WEIGHT: 1.0
50
+ NMS_CONFIG:
51
+ SCORE_THR: 0.01
52
+ NMS_PRE: 1000
53
+ IOU_THR: 0.5
54
+
55
+ ROI_HEAD:
56
+ NAME: CAGroup3DRoIHead
57
+ NUM_CLASSES: *N_CLASSES
58
+ MIDDLE_FEATURE_SOURCE: [3]
59
+ GRID_SIZE: 7
60
+ VOXEL_SIZE: *VOXEL_SIZE
61
+ COORD_KEY: 2
62
+ MLPS: [[64,128,128]]
63
+ CODE_SIZE: 6
64
+ ENCODE_SINCOS: False
65
+ ROI_PER_IMAGE: 128
66
+ ROI_FG_RATIO: 0.9
67
+ REG_FG_THRESH: 0.3
68
+ ROI_CONV_KERNEL: 5
69
+ ENLARGE_RATIO: False
70
+ USE_IOU_LOSS: False
71
+ USE_GRID_OFFSET: False
72
+ USE_SIMPLE_POOLING: True
73
+ USE_CENTER_POOLING: True
74
+ LOSS_WEIGHTS:
75
+ RCNN_CLS_WEIGHT: 1.0 # no use
76
+ RCNN_REG_WEIGHT: 1.0 # set to 0.5 if use iou loss
77
+ RCNN_IOU_WEIGHT: 1.0
78
+ CODE_WEIGHT: [1., 1., 1., 1., 1., 1.]
79
+
80
+ POST_PROCESSING:
81
+ RECALL_THRESH_LIST: [0.25, 0.5]
82
+ EVAL_METRIC: scannet
83
+
84
+ OPTIMIZATION:
85
+ BATCH_SIZE_PER_GPU: 16 # 4x4 or 8x2
86
+ NUM_EPOCHS: 1 #10
87
+
88
+ OPTIMIZER: adamW
89
+ LR: 0.001
90
+ WEIGHT_DECAY: 0.0001
91
+ DECAY_STEP_LIST: [7, 9]
92
+ LR_DECAY: 0.1
93
+ GRAD_NORM_CLIP: 10
94
+
95
+ # no use
96
+ PCT_START: 0.4
97
+ DIV_FACTOR: 10
98
+ LR_CLIP: 0.0000001
99
+
100
+ LR_WARMUP: False
101
+ WARMUP_EPOCH: 1
102
+
103
+
ckpt/checkpoint_epoch_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9eb4a927db7a3f710094f1e7b30317e7d4e95d6af8ab368946f7919d9455aa3f
3
+ size 1465028071
eval/eval_with_train/eval_list_val.txt ADDED
File without changes
eval/eval_with_train/tensorboard_val/events.out.tfevents.1680187773.DESKTOP-3FL13RB ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:710022cc5badae04a42df2e7816a28f1cb4533dddea6f3b43a9c2d5325fce8e0
3
+ size 40
log_train_20230326-130440.txt ADDED
@@ -0,0 +1,912 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2023-03-26 13:04:40,978 INFO **********************Start logging**********************
2
+ 2023-03-26 13:04:40,979 INFO CUDA_VISIBLE_DEVICES=ALL
3
+ 2023-03-26 13:04:40,980 INFO total_batch_size: 16
4
+ 2023-03-26 13:04:40,980 INFO cfg_file cfgs/scannet_models/CAGroup3D.yaml
5
+ 2023-03-26 13:04:40,981 INFO batch_size 16
6
+ 2023-03-26 13:04:40,981 INFO epochs 1
7
+ 2023-03-26 13:04:40,982 INFO workers 4
8
+ 2023-03-26 13:04:40,982 INFO extra_tag cagroup3d-win10-scannet
9
+ 2023-03-26 13:04:40,983 INFO ckpt None
10
+ 2023-03-26 13:04:40,984 INFO pretrained_model None
11
+ 2023-03-26 13:04:40,984 INFO launcher pytorch
12
+ 2023-03-26 13:04:40,985 INFO tcp_port 18888
13
+ 2023-03-26 13:04:40,985 INFO sync_bn False
14
+ 2023-03-26 13:04:40,986 INFO fix_random_seed True
15
+ 2023-03-26 13:04:40,986 INFO ckpt_save_interval 1
16
+ 2023-03-26 13:04:40,987 INFO max_ckpt_save_num 30
17
+ 2023-03-26 13:04:40,987 INFO merge_all_iters_to_one_epoch False
18
+ 2023-03-26 13:04:40,988 INFO set_cfgs None
19
+ 2023-03-26 13:04:40,988 INFO max_waiting_mins 0
20
+ 2023-03-26 13:04:40,989 INFO start_epoch 0
21
+ 2023-03-26 13:04:40,989 INFO num_epochs_to_eval 0
22
+ 2023-03-26 13:04:40,990 INFO save_to_file False
23
+ 2023-03-26 13:04:40,990 INFO cfg.ROOT_DIR: C:\CITYU\CS5182\proj\CAGroup3D
24
+ 2023-03-26 13:04:40,991 INFO cfg.LOCAL_RANK: 0
25
+ 2023-03-26 13:04:40,991 INFO cfg.CLASS_NAMES: ['cabinet', 'bed', 'chair', 'sofa', 'table', 'door', 'window', 'bookshelf', 'picture', 'counter', 'desk', 'curtain', 'refrigerator', 'showercurtrain', 'toilet', 'sink', 'bathtub', 'garbagebin']
26
+ 2023-03-26 13:04:40,992 INFO
27
+ cfg.DATA_CONFIG = edict()
28
+ 2023-03-26 13:04:40,993 INFO cfg.DATA_CONFIG.DATASET: ScannetDataset
29
+ 2023-03-26 13:04:40,993 INFO cfg.DATA_CONFIG.DATA_PATH: ../data/scannet_data/ScanNetV2
30
+ 2023-03-26 13:04:40,994 INFO cfg.DATA_CONFIG.PROCESSED_DATA_TAG: scannet_processed_data_v0_5_0
31
+ 2023-03-26 13:04:40,994 INFO cfg.DATA_CONFIG.POINT_CLOUD_RANGE: [-40, -40, -10, 40, 40, 10]
32
+ 2023-03-26 13:04:40,995 INFO
33
+ cfg.DATA_CONFIG.DATA_SPLIT = edict()
34
+ 2023-03-26 13:04:40,995 INFO cfg.DATA_CONFIG.DATA_SPLIT.train: train
35
+ 2023-03-26 13:04:40,996 INFO cfg.DATA_CONFIG.DATA_SPLIT.test: val
36
+ 2023-03-26 13:04:40,996 INFO
37
+ cfg.DATA_CONFIG.REPEAT = edict()
38
+ 2023-03-26 13:04:40,997 INFO cfg.DATA_CONFIG.REPEAT.train: 10
39
+ 2023-03-26 13:04:40,998 INFO cfg.DATA_CONFIG.REPEAT.test: 1
40
+ 2023-03-26 13:04:40,998 INFO
41
+ cfg.DATA_CONFIG.INFO_PATH = edict()
42
+ 2023-03-26 13:04:40,999 INFO cfg.DATA_CONFIG.INFO_PATH.train: ['scannet_infos_train.pkl']
43
+ 2023-03-26 13:04:40,999 INFO cfg.DATA_CONFIG.INFO_PATH.test: ['scannet_infos_val.pkl']
44
+ 2023-03-26 13:04:41,000 INFO cfg.DATA_CONFIG.GET_ITEM_LIST: ['points', 'instance_mask', 'semantic_mask']
45
+ 2023-03-26 13:04:41,000 INFO cfg.DATA_CONFIG.FILTER_EMPTY_BOXES_FOR_TRAIN: True
46
+ 2023-03-26 13:04:41,001 INFO
47
+ cfg.DATA_CONFIG.DATA_AUGMENTOR_TRAIN = edict()
48
+ 2023-03-26 13:04:41,002 INFO cfg.DATA_CONFIG.DATA_AUGMENTOR_TRAIN.DISABLE_AUG_LIST: ['placeholder']
49
+ 2023-03-26 13:04:41,003 INFO cfg.DATA_CONFIG.DATA_AUGMENTOR_TRAIN.AUG_CONFIG_LIST: [{'NAME': 'global_alignment', 'rotation_axis': 2}, {'NAME': 'point_seg_class_mapping', 'valid_cat_ids': [3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 14, 16, 24, 28, 33, 34, 36, 39], 'max_cat_id': 40}, {'NAME': 'random_world_flip', 'ALONG_AXIS_LIST': ['x', 'y']}, {'NAME': 'random_world_rotation', 'WORLD_ROT_ANGLE': [-0.087266, 0.087266]}, {'NAME': 'random_world_scaling', 'WORLD_SCALE_RANGE': [0.9, 1.1]}, {'NAME': 'random_world_translation', 'ALONG_AXIS_LIST': ['x', 'y', 'z'], 'NOISE_TRANSLATE_STD': 0.1}]
50
+ 2023-03-26 13:04:41,004 INFO
51
+ cfg.DATA_CONFIG.DATA_AUGMENTOR_TEST = edict()
52
+ 2023-03-26 13:04:41,004 INFO cfg.DATA_CONFIG.DATA_AUGMENTOR_TEST.DISABLE_AUG_LIST: ['placeholder']
53
+ 2023-03-26 13:04:41,005 INFO cfg.DATA_CONFIG.DATA_AUGMENTOR_TEST.AUG_CONFIG_LIST: [{'NAME': 'global_alignment', 'rotation_axis': 2}]
54
+ 2023-03-26 13:04:41,005 INFO
55
+ cfg.DATA_CONFIG.DATA_AUGMENTOR = edict()
56
+ 2023-03-26 13:04:41,006 INFO cfg.DATA_CONFIG.DATA_AUGMENTOR.DISABLE_AUG_LIST: ['placeholder']
57
+ 2023-03-26 13:04:41,007 INFO cfg.DATA_CONFIG.DATA_AUGMENTOR.AUG_CONFIG_LIST: [{'NAME': 'global_alignment', 'rotation_axis': 2}]
58
+ 2023-03-26 13:04:41,008 INFO
59
+ cfg.DATA_CONFIG.POINT_FEATURE_ENCODING = edict()
60
+ 2023-03-26 13:04:41,008 INFO cfg.DATA_CONFIG.POINT_FEATURE_ENCODING.encoding_type: absolute_coordinates_encoding
61
+ 2023-03-26 13:04:41,009 INFO cfg.DATA_CONFIG.POINT_FEATURE_ENCODING.used_feature_list: ['x', 'y', 'z', 'r', 'g', 'b']
62
+ 2023-03-26 13:04:41,009 INFO cfg.DATA_CONFIG.POINT_FEATURE_ENCODING.src_feature_list: ['x', 'y', 'z', 'r', 'g', 'b']
63
+ 2023-03-26 13:04:41,010 INFO cfg.DATA_CONFIG.DATA_PROCESSOR: [{'NAME': 'mask_points_and_boxes_outside_range', 'REMOVE_OUTSIDE_BOXES': True}]
64
+ 2023-03-26 13:04:41,010 INFO cfg.DATA_CONFIG._BASE_CONFIG_: cfgs/dataset_configs/scannet_dataset.yaml
65
+ 2023-03-26 13:04:41,011 INFO cfg.VOXEL_SIZE: 0.02
66
+ 2023-03-26 13:04:41,011 INFO cfg.N_CLASSES: 18
67
+ 2023-03-26 13:04:41,012 INFO cfg.SEMANTIC_THR: 0.15
68
+ 2023-03-26 13:04:41,012 INFO
69
+ cfg.MODEL = edict()
70
+ 2023-03-26 13:04:41,013 INFO cfg.MODEL.NAME: CAGroup3D
71
+ 2023-03-26 13:04:41,013 INFO cfg.MODEL.VOXEL_SIZE: 0.02
72
+ 2023-03-26 13:04:41,013 INFO cfg.MODEL.SEMANTIC_MIN_THR: 0.05
73
+ 2023-03-26 13:04:41,014 INFO cfg.MODEL.SEMANTIC_ITER_VALUE: 0.02
74
+ 2023-03-26 13:04:41,014 INFO cfg.MODEL.SEMANTIC_THR: 0.15
75
+ 2023-03-26 13:04:41,015 INFO
76
+ cfg.MODEL.BACKBONE_3D = edict()
77
+ 2023-03-26 13:04:41,015 INFO cfg.MODEL.BACKBONE_3D.NAME: BiResNet
78
+ 2023-03-26 13:04:41,016 INFO cfg.MODEL.BACKBONE_3D.IN_CHANNELS: 3
79
+ 2023-03-26 13:04:41,016 INFO cfg.MODEL.BACKBONE_3D.OUT_CHANNELS: 64
80
+ 2023-03-26 13:04:41,017 INFO
81
+ cfg.MODEL.DENSE_HEAD = edict()
82
+ 2023-03-26 13:04:41,017 INFO cfg.MODEL.DENSE_HEAD.NAME: CAGroup3DHead
83
+ 2023-03-26 13:04:41,018 INFO cfg.MODEL.DENSE_HEAD.IN_CHANNELS: [64, 128, 256, 512]
84
+ 2023-03-26 13:04:41,018 INFO cfg.MODEL.DENSE_HEAD.OUT_CHANNELS: 64
85
+ 2023-03-26 13:04:41,019 INFO cfg.MODEL.DENSE_HEAD.SEMANTIC_THR: 0.15
86
+ 2023-03-26 13:04:41,019 INFO cfg.MODEL.DENSE_HEAD.VOXEL_SIZE: 0.02
87
+ 2023-03-26 13:04:41,020 INFO cfg.MODEL.DENSE_HEAD.N_CLASSES: 18
88
+ 2023-03-26 13:04:41,020 INFO cfg.MODEL.DENSE_HEAD.N_REG_OUTS: 6
89
+ 2023-03-26 13:04:41,021 INFO cfg.MODEL.DENSE_HEAD.CLS_KERNEL: 9
90
+ 2023-03-26 13:04:41,021 INFO cfg.MODEL.DENSE_HEAD.WITH_YAW: False
91
+ 2023-03-26 13:04:41,022 INFO cfg.MODEL.DENSE_HEAD.USE_SEM_SCORE: False
92
+ 2023-03-26 13:04:41,022 INFO cfg.MODEL.DENSE_HEAD.EXPAND_RATIO: 3
93
+ 2023-03-26 13:04:41,022 INFO
94
+ cfg.MODEL.DENSE_HEAD.ASSIGNER = edict()
95
+ 2023-03-26 13:04:41,023 INFO cfg.MODEL.DENSE_HEAD.ASSIGNER.NAME: CAGroup3DAssigner
96
+ 2023-03-26 13:04:41,023 INFO cfg.MODEL.DENSE_HEAD.ASSIGNER.LIMIT: 27
97
+ 2023-03-26 13:04:41,024 INFO cfg.MODEL.DENSE_HEAD.ASSIGNER.TOPK: 18
98
+ 2023-03-26 13:04:41,024 INFO cfg.MODEL.DENSE_HEAD.ASSIGNER.N_SCALES: 4
99
+ 2023-03-26 13:04:41,025 INFO
100
+ cfg.MODEL.DENSE_HEAD.LOSS_OFFSET = edict()
101
+ 2023-03-26 13:04:41,025 INFO cfg.MODEL.DENSE_HEAD.LOSS_OFFSET.NAME: SmoothL1Loss
102
+ 2023-03-26 13:04:41,026 INFO cfg.MODEL.DENSE_HEAD.LOSS_OFFSET.BETA: 0.04
103
+ 2023-03-26 13:04:41,026 INFO cfg.MODEL.DENSE_HEAD.LOSS_OFFSET.REDUCTION: sum
104
+ 2023-03-26 13:04:41,027 INFO cfg.MODEL.DENSE_HEAD.LOSS_OFFSET.LOSS_WEIGHT: 1.0
105
+ 2023-03-26 13:04:41,027 INFO
106
+ cfg.MODEL.DENSE_HEAD.LOSS_BBOX = edict()
107
+ 2023-03-26 13:04:41,028 INFO cfg.MODEL.DENSE_HEAD.LOSS_BBOX.NAME: IoU3DLoss
108
+ 2023-03-26 13:04:41,028 INFO cfg.MODEL.DENSE_HEAD.LOSS_BBOX.WITH_YAW: False
109
+ 2023-03-26 13:04:41,028 INFO cfg.MODEL.DENSE_HEAD.LOSS_BBOX.LOSS_WEIGHT: 1.0
110
+ 2023-03-26 13:04:41,029 INFO
111
+ cfg.MODEL.DENSE_HEAD.NMS_CONFIG = edict()
112
+ 2023-03-26 13:04:41,029 INFO cfg.MODEL.DENSE_HEAD.NMS_CONFIG.SCORE_THR: 0.01
113
+ 2023-03-26 13:04:41,030 INFO cfg.MODEL.DENSE_HEAD.NMS_CONFIG.NMS_PRE: 1000
114
+ 2023-03-26 13:04:41,030 INFO cfg.MODEL.DENSE_HEAD.NMS_CONFIG.IOU_THR: 0.5
115
+ 2023-03-26 13:04:41,031 INFO
116
+ cfg.MODEL.ROI_HEAD = edict()
117
+ 2023-03-26 13:04:41,031 INFO cfg.MODEL.ROI_HEAD.NAME: CAGroup3DRoIHead
118
+ 2023-03-26 13:04:41,032 INFO cfg.MODEL.ROI_HEAD.NUM_CLASSES: 18
119
+ 2023-03-26 13:04:41,032 INFO cfg.MODEL.ROI_HEAD.MIDDLE_FEATURE_SOURCE: [3]
120
+ 2023-03-26 13:04:41,033 INFO cfg.MODEL.ROI_HEAD.GRID_SIZE: 7
121
+ 2023-03-26 13:04:41,033 INFO cfg.MODEL.ROI_HEAD.VOXEL_SIZE: 0.02
122
+ 2023-03-26 13:04:41,034 INFO cfg.MODEL.ROI_HEAD.COORD_KEY: 2
123
+ 2023-03-26 13:04:41,034 INFO cfg.MODEL.ROI_HEAD.MLPS: [[64, 128, 128]]
124
+ 2023-03-26 13:04:41,035 INFO cfg.MODEL.ROI_HEAD.CODE_SIZE: 6
125
+ 2023-03-26 13:04:41,035 INFO cfg.MODEL.ROI_HEAD.ENCODE_SINCOS: False
126
+ 2023-03-26 13:04:41,036 INFO cfg.MODEL.ROI_HEAD.ROI_PER_IMAGE: 128
127
+ 2023-03-26 13:04:41,036 INFO cfg.MODEL.ROI_HEAD.ROI_FG_RATIO: 0.9
128
+ 2023-03-26 13:04:41,036 INFO cfg.MODEL.ROI_HEAD.REG_FG_THRESH: 0.3
129
+ 2023-03-26 13:04:41,037 INFO cfg.MODEL.ROI_HEAD.ROI_CONV_KERNEL: 5
130
+ 2023-03-26 13:04:41,037 INFO cfg.MODEL.ROI_HEAD.ENLARGE_RATIO: False
131
+ 2023-03-26 13:04:41,038 INFO cfg.MODEL.ROI_HEAD.USE_IOU_LOSS: False
132
+ 2023-03-26 13:04:41,038 INFO cfg.MODEL.ROI_HEAD.USE_GRID_OFFSET: False
133
+ 2023-03-26 13:04:41,039 INFO cfg.MODEL.ROI_HEAD.USE_SIMPLE_POOLING: True
134
+ 2023-03-26 13:04:41,039 INFO cfg.MODEL.ROI_HEAD.USE_CENTER_POOLING: True
135
+ 2023-03-26 13:04:41,039 INFO
136
+ cfg.MODEL.ROI_HEAD.LOSS_WEIGHTS = edict()
137
+ 2023-03-26 13:04:41,040 INFO cfg.MODEL.ROI_HEAD.LOSS_WEIGHTS.RCNN_CLS_WEIGHT: 1.0
138
+ 2023-03-26 13:04:41,040 INFO cfg.MODEL.ROI_HEAD.LOSS_WEIGHTS.RCNN_REG_WEIGHT: 1.0
139
+ 2023-03-26 13:04:41,041 INFO cfg.MODEL.ROI_HEAD.LOSS_WEIGHTS.RCNN_IOU_WEIGHT: 1.0
140
+ 2023-03-26 13:04:41,041 INFO cfg.MODEL.ROI_HEAD.LOSS_WEIGHTS.CODE_WEIGHT: [1.0, 1.0, 1.0, 1.0, 1.0, 1.0]
141
+ 2023-03-26 13:04:41,042 INFO
142
+ cfg.MODEL.POST_PROCESSING = edict()
143
+ 2023-03-26 13:04:41,042 INFO cfg.MODEL.POST_PROCESSING.RECALL_THRESH_LIST: [0.25, 0.5]
144
+ 2023-03-26 13:04:41,043 INFO cfg.MODEL.POST_PROCESSING.EVAL_METRIC: scannet
145
+ 2023-03-26 13:04:41,043 INFO
146
+ cfg.OPTIMIZATION = edict()
147
+ 2023-03-26 13:04:41,044 INFO cfg.OPTIMIZATION.BATCH_SIZE_PER_GPU: 16
148
+ 2023-03-26 13:04:41,044 INFO cfg.OPTIMIZATION.NUM_EPOCHS: 1
149
+ 2023-03-26 13:04:41,044 INFO cfg.OPTIMIZATION.OPTIMIZER: adamW
150
+ 2023-03-26 13:04:41,045 INFO cfg.OPTIMIZATION.LR: 0.001
151
+ 2023-03-26 13:04:41,045 INFO cfg.OPTIMIZATION.WEIGHT_DECAY: 0.0001
152
+ 2023-03-26 13:04:41,046 INFO cfg.OPTIMIZATION.DECAY_STEP_LIST: [7, 9]
153
+ 2023-03-26 13:04:41,046 INFO cfg.OPTIMIZATION.LR_DECAY: 0.1
154
+ 2023-03-26 13:04:41,046 INFO cfg.OPTIMIZATION.GRAD_NORM_CLIP: 10
155
+ 2023-03-26 13:04:41,047 INFO cfg.OPTIMIZATION.PCT_START: 0.4
156
+ 2023-03-26 13:04:41,047 INFO cfg.OPTIMIZATION.DIV_FACTOR: 10
157
+ 2023-03-26 13:04:41,048 INFO cfg.OPTIMIZATION.LR_CLIP: 1e-07
158
+ 2023-03-26 13:04:41,048 INFO cfg.OPTIMIZATION.LR_WARMUP: False
159
+ 2023-03-26 13:04:41,049 INFO cfg.OPTIMIZATION.WARMUP_EPOCH: 1
160
+ 2023-03-26 13:04:41,049 INFO cfg.TAG: CAGroup3D
161
+ 2023-03-26 13:04:41,049 INFO cfg.EXP_GROUP_PATH: scannet_models
162
+ 2023-03-26 13:04:41,085 INFO Loading SCANNET dataset
163
+ 2023-03-26 13:04:41,192 INFO Total samples for SCANNET dataset: 1201
164
+ 2023-03-26 13:04:44,269 INFO DistributedDataParallel(
165
+ (module): CAGroup3D(
166
+ (vfe): None
167
+ (backbone_3d): BiResNet(
168
+ (conv1): Sequential(
169
+ (0): MinkowskiConvolution(in=3, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
170
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
171
+ (2): MinkowskiReLU()
172
+ (3): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
173
+ (4): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
174
+ (5): MinkowskiReLU()
175
+ )
176
+ (relu): MinkowskiReLU()
177
+ (layer1): Sequential(
178
+ (0): BasicBlock(
179
+ (conv1): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[2, 2, 2], dilation=[1, 1, 1])
180
+ (norm1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
181
+ (conv2): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
182
+ (norm2): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
183
+ (relu): MinkowskiReLU()
184
+ (downsample): Sequential(
185
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[1, 1, 1], stride=[2, 2, 2], dilation=[1, 1, 1])
186
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
187
+ )
188
+ )
189
+ (1): BasicBlock(
190
+ (conv1): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
191
+ (norm1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
192
+ (conv2): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
193
+ (norm2): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
194
+ (relu): MinkowskiReLU()
195
+ )
196
+ )
197
+ (layer2): Sequential(
198
+ (0): BasicBlock(
199
+ (conv1): MinkowskiConvolution(in=64, out=128, kernel_size=[3, 3, 3], stride=[2, 2, 2], dilation=[1, 1, 1])
200
+ (norm1): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
201
+ (conv2): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
202
+ (norm2): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
203
+ (relu): MinkowskiReLU()
204
+ (downsample): Sequential(
205
+ (0): MinkowskiConvolution(in=64, out=128, kernel_size=[1, 1, 1], stride=[2, 2, 2], dilation=[1, 1, 1])
206
+ (1): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
207
+ )
208
+ )
209
+ (1): BasicBlock(
210
+ (conv1): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
211
+ (norm1): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
212
+ (conv2): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
213
+ (norm2): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
214
+ (relu): MinkowskiReLU()
215
+ )
216
+ )
217
+ (layer3): Sequential(
218
+ (0): BasicBlock(
219
+ (conv1): MinkowskiConvolution(in=128, out=256, kernel_size=[3, 3, 3], stride=[2, 2, 2], dilation=[1, 1, 1])
220
+ (norm1): MinkowskiBatchNorm(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
221
+ (conv2): MinkowskiConvolution(in=256, out=256, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
222
+ (norm2): MinkowskiBatchNorm(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
223
+ (relu): MinkowskiReLU()
224
+ (downsample): Sequential(
225
+ (0): MinkowskiConvolution(in=128, out=256, kernel_size=[1, 1, 1], stride=[2, 2, 2], dilation=[1, 1, 1])
226
+ (1): MinkowskiBatchNorm(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
227
+ )
228
+ )
229
+ (1): BasicBlock(
230
+ (conv1): MinkowskiConvolution(in=256, out=256, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
231
+ (norm1): MinkowskiBatchNorm(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
232
+ (conv2): MinkowskiConvolution(in=256, out=256, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
233
+ (norm2): MinkowskiBatchNorm(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
234
+ (relu): MinkowskiReLU()
235
+ )
236
+ )
237
+ (layer4): Sequential(
238
+ (0): BasicBlock(
239
+ (conv1): MinkowskiConvolution(in=256, out=512, kernel_size=[3, 3, 3], stride=[2, 2, 2], dilation=[1, 1, 1])
240
+ (norm1): MinkowskiBatchNorm(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
241
+ (conv2): MinkowskiConvolution(in=512, out=512, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
242
+ (norm2): MinkowskiBatchNorm(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
243
+ (relu): MinkowskiReLU()
244
+ (downsample): Sequential(
245
+ (0): MinkowskiConvolution(in=256, out=512, kernel_size=[1, 1, 1], stride=[2, 2, 2], dilation=[1, 1, 1])
246
+ (1): MinkowskiBatchNorm(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
247
+ )
248
+ )
249
+ (1): BasicBlock(
250
+ (conv1): MinkowskiConvolution(in=512, out=512, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
251
+ (norm1): MinkowskiBatchNorm(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
252
+ (conv2): MinkowskiConvolution(in=512, out=512, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
253
+ (norm2): MinkowskiBatchNorm(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
254
+ (relu): MinkowskiReLU()
255
+ )
256
+ )
257
+ (compression3): Sequential(
258
+ (0): MinkowskiConvolution(in=256, out=128, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
259
+ (1): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
260
+ )
261
+ (compression4): Sequential(
262
+ (0): MinkowskiConvolution(in=512, out=128, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
263
+ (1): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
264
+ )
265
+ (down3): Sequential(
266
+ (0): MinkowskiConvolution(in=128, out=256, kernel_size=[3, 3, 3], stride=[2, 2, 2], dilation=[1, 1, 1])
267
+ (1): MinkowskiBatchNorm(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
268
+ )
269
+ (down4): Sequential(
270
+ (0): MinkowskiConvolution(in=128, out=256, kernel_size=[3, 3, 3], stride=[2, 2, 2], dilation=[1, 1, 1])
271
+ (1): MinkowskiBatchNorm(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
272
+ (2): MinkowskiReLU()
273
+ (3): MinkowskiConvolution(in=256, out=512, kernel_size=[3, 3, 3], stride=[2, 2, 2], dilation=[1, 1, 1])
274
+ (4): MinkowskiBatchNorm(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
275
+ )
276
+ (layer3_): Sequential(
277
+ (0): BasicBlock(
278
+ (conv1): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
279
+ (norm1): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
280
+ (conv2): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
281
+ (norm2): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
282
+ (relu): MinkowskiReLU()
283
+ )
284
+ (1): BasicBlock(
285
+ (conv1): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
286
+ (norm1): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
287
+ (conv2): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
288
+ (norm2): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
289
+ (relu): MinkowskiReLU()
290
+ )
291
+ )
292
+ (layer4_): Sequential(
293
+ (0): BasicBlock(
294
+ (conv1): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
295
+ (norm1): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
296
+ (conv2): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
297
+ (norm2): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
298
+ (relu): MinkowskiReLU()
299
+ )
300
+ (1): BasicBlock(
301
+ (conv1): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
302
+ (norm1): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
303
+ (conv2): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
304
+ (norm2): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
305
+ (relu): MinkowskiReLU()
306
+ )
307
+ )
308
+ (layer5_): Sequential(
309
+ (0): Bottleneck(
310
+ (conv1): MinkowskiConvolution(in=128, out=128, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
311
+ (norm1): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
312
+ (conv2): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
313
+ (norm2): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
314
+ (conv3): MinkowskiConvolution(in=128, out=256, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
315
+ (norm3): MinkowskiBatchNorm(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
316
+ (relu): MinkowskiReLU()
317
+ (downsample): Sequential(
318
+ (0): MinkowskiConvolution(in=128, out=256, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
319
+ (1): MinkowskiBatchNorm(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
320
+ )
321
+ )
322
+ )
323
+ (layer5): Sequential(
324
+ (0): Bottleneck(
325
+ (conv1): MinkowskiConvolution(in=512, out=512, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
326
+ (norm1): MinkowskiBatchNorm(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
327
+ (conv2): MinkowskiConvolution(in=512, out=512, kernel_size=[3, 3, 3], stride=[2, 2, 2], dilation=[1, 1, 1])
328
+ (norm2): MinkowskiBatchNorm(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
329
+ (conv3): MinkowskiConvolution(in=512, out=1024, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
330
+ (norm3): MinkowskiBatchNorm(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
331
+ (relu): MinkowskiReLU()
332
+ (downsample): Sequential(
333
+ (0): MinkowskiConvolution(in=512, out=1024, kernel_size=[1, 1, 1], stride=[2, 2, 2], dilation=[1, 1, 1])
334
+ (1): MinkowskiBatchNorm(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
335
+ )
336
+ )
337
+ )
338
+ (spp): DAPPM(
339
+ (scale1): Sequential(
340
+ (0): MinkowskiAvgPooling(kernel_size=[5, 5, 5], stride=[2, 2, 2], dilation=[1, 1, 1])
341
+ (1): MinkowskiBatchNorm(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
342
+ (2): MinkowskiReLU()
343
+ (3): MinkowskiConvolution(in=1024, out=128, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
344
+ )
345
+ (scale2): Sequential(
346
+ (0): MinkowskiAvgPooling(kernel_size=[9, 9, 9], stride=[4, 4, 4], dilation=[1, 1, 1])
347
+ (1): MinkowskiBatchNorm(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
348
+ (2): MinkowskiReLU()
349
+ (3): MinkowskiConvolution(in=1024, out=128, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
350
+ )
351
+ (scale3): Sequential(
352
+ (0): MinkowskiAvgPooling(kernel_size=[17, 17, 17], stride=[8, 8, 8], dilation=[1, 1, 1])
353
+ (1): MinkowskiBatchNorm(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
354
+ (2): MinkowskiReLU()
355
+ (3): MinkowskiConvolution(in=1024, out=128, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
356
+ )
357
+ (scale4): Sequential(
358
+ (0): MinkowskiAvgPooling(kernel_size=[33, 33, 33], stride=[16, 16, 16], dilation=[1, 1, 1])
359
+ (1): MinkowskiBatchNorm(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
360
+ (2): MinkowskiReLU()
361
+ (3): MinkowskiConvolution(in=1024, out=128, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
362
+ )
363
+ (scale0): Sequential(
364
+ (0): MinkowskiBatchNorm(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
365
+ (1): MinkowskiReLU()
366
+ (2): MinkowskiConvolution(in=1024, out=128, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
367
+ )
368
+ (process1): Sequential(
369
+ (0): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
370
+ (1): MinkowskiReLU()
371
+ (2): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
372
+ )
373
+ (process2): Sequential(
374
+ (0): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
375
+ (1): MinkowskiReLU()
376
+ (2): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
377
+ )
378
+ (process3): Sequential(
379
+ (0): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
380
+ (1): MinkowskiReLU()
381
+ (2): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
382
+ )
383
+ (process4): Sequential(
384
+ (0): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
385
+ (1): MinkowskiReLU()
386
+ (2): MinkowskiConvolution(in=128, out=128, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
387
+ )
388
+ (compression): Sequential(
389
+ (0): MinkowskiBatchNorm(640, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
390
+ (1): MinkowskiReLU()
391
+ (2): MinkowskiConvolution(in=640, out=256, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
392
+ )
393
+ (shortcut): Sequential(
394
+ (0): MinkowskiBatchNorm(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
395
+ (1): MinkowskiReLU()
396
+ (2): MinkowskiConvolution(in=1024, out=256, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
397
+ )
398
+ )
399
+ (out): Sequential(
400
+ (0): MinkowskiConvolutionTranspose(in=256, out=256, kernel_size=[2, 2, 2], stride=[2, 2, 2], dilation=[1, 1, 1])
401
+ (1): MinkowskiBatchNorm(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
402
+ (2): MinkowskiReLU()
403
+ (3): MinkowskiConvolution(in=256, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
404
+ (4): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
405
+ (5): MinkowskiReLU()
406
+ )
407
+ )
408
+ (map_to_bev_module): None
409
+ (pfe): None
410
+ (backbone_2d): None
411
+ (dense_head): CAGroup3DHead(
412
+ (loss_centerness): CrossEntropy()
413
+ (loss_bbox): IoU3DLoss()
414
+ (loss_cls): FocalLoss()
415
+ (loss_sem): FocalLoss()
416
+ (loss_offset): SmoothL1Loss()
417
+ (offset_block): Sequential(
418
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
419
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
420
+ (2): MinkowskiELU()
421
+ (3): MinkowskiConvolution(in=64, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
422
+ (4): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
423
+ (5): MinkowskiELU()
424
+ (6): MinkowskiConvolution(in=64, out=3, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
425
+ )
426
+ (feature_offset): Sequential(
427
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
428
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
429
+ (2): MinkowskiELU()
430
+ )
431
+ (semantic_conv): MinkowskiConvolution(in=64, out=18, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
432
+ (centerness_conv): MinkowskiConvolution(in=64, out=1, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
433
+ (reg_conv): MinkowskiConvolution(in=64, out=6, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
434
+ (cls_conv): MinkowskiConvolution(in=64, out=18, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
435
+ (scales): ModuleList(
436
+ (0): Scale()
437
+ (1): Scale()
438
+ (2): Scale()
439
+ (3): Scale()
440
+ (4): Scale()
441
+ (5): Scale()
442
+ (6): Scale()
443
+ (7): Scale()
444
+ (8): Scale()
445
+ (9): Scale()
446
+ (10): Scale()
447
+ (11): Scale()
448
+ (12): Scale()
449
+ (13): Scale()
450
+ (14): Scale()
451
+ (15): Scale()
452
+ (16): Scale()
453
+ (17): Scale()
454
+ )
455
+ (cls_individual_out): ModuleList(
456
+ (0): Sequential(
457
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
458
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
459
+ (2): MinkowskiELU()
460
+ )
461
+ (1): Sequential(
462
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
463
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
464
+ (2): MinkowskiELU()
465
+ )
466
+ (2): Sequential(
467
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
468
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
469
+ (2): MinkowskiELU()
470
+ )
471
+ (3): Sequential(
472
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
473
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
474
+ (2): MinkowskiELU()
475
+ )
476
+ (4): Sequential(
477
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
478
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
479
+ (2): MinkowskiELU()
480
+ )
481
+ (5): Sequential(
482
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
483
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
484
+ (2): MinkowskiELU()
485
+ )
486
+ (6): Sequential(
487
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
488
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
489
+ (2): MinkowskiELU()
490
+ )
491
+ (7): Sequential(
492
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
493
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
494
+ (2): MinkowskiELU()
495
+ )
496
+ (8): Sequential(
497
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
498
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
499
+ (2): MinkowskiELU()
500
+ )
501
+ (9): Sequential(
502
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
503
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
504
+ (2): MinkowskiELU()
505
+ )
506
+ (10): Sequential(
507
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
508
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
509
+ (2): MinkowskiELU()
510
+ )
511
+ (11): Sequential(
512
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
513
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
514
+ (2): MinkowskiELU()
515
+ )
516
+ (12): Sequential(
517
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
518
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
519
+ (2): MinkowskiELU()
520
+ )
521
+ (13): Sequential(
522
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
523
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
524
+ (2): MinkowskiELU()
525
+ )
526
+ (14): Sequential(
527
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
528
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
529
+ (2): MinkowskiELU()
530
+ )
531
+ (15): Sequential(
532
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
533
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
534
+ (2): MinkowskiELU()
535
+ )
536
+ (16): Sequential(
537
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
538
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
539
+ (2): MinkowskiELU()
540
+ )
541
+ (17): Sequential(
542
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[9, 9, 9], stride=[1, 1, 1], dilation=[1, 1, 1])
543
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
544
+ (2): MinkowskiELU()
545
+ )
546
+ )
547
+ (cls_individual_up): ModuleList(
548
+ (0): ModuleList(
549
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
550
+ (1): Sequential(
551
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
552
+ (1): MinkowskiELU()
553
+ )
554
+ )
555
+ (1): ModuleList(
556
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
557
+ (1): Sequential(
558
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
559
+ (1): MinkowskiELU()
560
+ )
561
+ )
562
+ (2): ModuleList(
563
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
564
+ (1): Sequential(
565
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
566
+ (1): MinkowskiELU()
567
+ )
568
+ )
569
+ (3): ModuleList(
570
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
571
+ (1): Sequential(
572
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
573
+ (1): MinkowskiELU()
574
+ )
575
+ )
576
+ (4): ModuleList(
577
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
578
+ (1): Sequential(
579
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
580
+ (1): MinkowskiELU()
581
+ )
582
+ )
583
+ (5): ModuleList(
584
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
585
+ (1): Sequential(
586
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
587
+ (1): MinkowskiELU()
588
+ )
589
+ )
590
+ (6): ModuleList(
591
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
592
+ (1): Sequential(
593
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
594
+ (1): MinkowskiELU()
595
+ )
596
+ )
597
+ (7): ModuleList(
598
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
599
+ (1): Sequential(
600
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
601
+ (1): MinkowskiELU()
602
+ )
603
+ )
604
+ (8): ModuleList(
605
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
606
+ (1): Sequential(
607
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
608
+ (1): MinkowskiELU()
609
+ )
610
+ )
611
+ (9): ModuleList(
612
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
613
+ (1): Sequential(
614
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
615
+ (1): MinkowskiELU()
616
+ )
617
+ )
618
+ (10): ModuleList(
619
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
620
+ (1): Sequential(
621
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
622
+ (1): MinkowskiELU()
623
+ )
624
+ )
625
+ (11): ModuleList(
626
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
627
+ (1): Sequential(
628
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
629
+ (1): MinkowskiELU()
630
+ )
631
+ )
632
+ (12): ModuleList(
633
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
634
+ (1): Sequential(
635
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
636
+ (1): MinkowskiELU()
637
+ )
638
+ )
639
+ (13): ModuleList(
640
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
641
+ (1): Sequential(
642
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
643
+ (1): MinkowskiELU()
644
+ )
645
+ )
646
+ (14): ModuleList(
647
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
648
+ (1): Sequential(
649
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
650
+ (1): MinkowskiELU()
651
+ )
652
+ )
653
+ (15): ModuleList(
654
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
655
+ (1): Sequential(
656
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
657
+ (1): MinkowskiELU()
658
+ )
659
+ )
660
+ (16): ModuleList(
661
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
662
+ (1): Sequential(
663
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
664
+ (1): MinkowskiELU()
665
+ )
666
+ )
667
+ (17): ModuleList(
668
+ (0): MinkowskiGenerativeConvolutionTranspose(in=64, out=64, kernel_size=[3, 3, 3], stride=[3, 3, 3], dilation=[1, 1, 1])
669
+ (1): Sequential(
670
+ (0): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
671
+ (1): MinkowskiELU()
672
+ )
673
+ )
674
+ )
675
+ (cls_individual_fuse): ModuleList(
676
+ (0): Sequential(
677
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
678
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
679
+ (2): MinkowskiELU()
680
+ )
681
+ (1): Sequential(
682
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
683
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
684
+ (2): MinkowskiELU()
685
+ )
686
+ (2): Sequential(
687
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
688
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
689
+ (2): MinkowskiELU()
690
+ )
691
+ (3): Sequential(
692
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
693
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
694
+ (2): MinkowskiELU()
695
+ )
696
+ (4): Sequential(
697
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
698
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
699
+ (2): MinkowskiELU()
700
+ )
701
+ (5): Sequential(
702
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
703
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
704
+ (2): MinkowskiELU()
705
+ )
706
+ (6): Sequential(
707
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
708
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
709
+ (2): MinkowskiELU()
710
+ )
711
+ (7): Sequential(
712
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
713
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
714
+ (2): MinkowskiELU()
715
+ )
716
+ (8): Sequential(
717
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
718
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
719
+ (2): MinkowskiELU()
720
+ )
721
+ (9): Sequential(
722
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
723
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
724
+ (2): MinkowskiELU()
725
+ )
726
+ (10): Sequential(
727
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
728
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
729
+ (2): MinkowskiELU()
730
+ )
731
+ (11): Sequential(
732
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
733
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
734
+ (2): MinkowskiELU()
735
+ )
736
+ (12): Sequential(
737
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
738
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
739
+ (2): MinkowskiELU()
740
+ )
741
+ (13): Sequential(
742
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
743
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
744
+ (2): MinkowskiELU()
745
+ )
746
+ (14): Sequential(
747
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
748
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
749
+ (2): MinkowskiELU()
750
+ )
751
+ (15): Sequential(
752
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
753
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
754
+ (2): MinkowskiELU()
755
+ )
756
+ (16): Sequential(
757
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
758
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
759
+ (2): MinkowskiELU()
760
+ )
761
+ (17): Sequential(
762
+ (0): MinkowskiConvolution(in=128, out=64, kernel_size=[1, 1, 1], stride=[1, 1, 1], dilation=[1, 1, 1])
763
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
764
+ (2): MinkowskiELU()
765
+ )
766
+ )
767
+ (cls_individual_expand_out): ModuleList(
768
+ (0): Sequential(
769
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
770
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
771
+ (2): MinkowskiELU()
772
+ )
773
+ (1): Sequential(
774
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
775
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
776
+ (2): MinkowskiELU()
777
+ )
778
+ (2): Sequential(
779
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
780
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
781
+ (2): MinkowskiELU()
782
+ )
783
+ (3): Sequential(
784
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
785
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
786
+ (2): MinkowskiELU()
787
+ )
788
+ (4): Sequential(
789
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
790
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
791
+ (2): MinkowskiELU()
792
+ )
793
+ (5): Sequential(
794
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
795
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
796
+ (2): MinkowskiELU()
797
+ )
798
+ (6): Sequential(
799
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
800
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
801
+ (2): MinkowskiELU()
802
+ )
803
+ (7): Sequential(
804
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
805
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
806
+ (2): MinkowskiELU()
807
+ )
808
+ (8): Sequential(
809
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
810
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
811
+ (2): MinkowskiELU()
812
+ )
813
+ (9): Sequential(
814
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
815
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
816
+ (2): MinkowskiELU()
817
+ )
818
+ (10): Sequential(
819
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
820
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
821
+ (2): MinkowskiELU()
822
+ )
823
+ (11): Sequential(
824
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
825
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
826
+ (2): MinkowskiELU()
827
+ )
828
+ (12): Sequential(
829
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
830
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
831
+ (2): MinkowskiELU()
832
+ )
833
+ (13): Sequential(
834
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
835
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
836
+ (2): MinkowskiELU()
837
+ )
838
+ (14): Sequential(
839
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
840
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
841
+ (2): MinkowskiELU()
842
+ )
843
+ (15): Sequential(
844
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
845
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
846
+ (2): MinkowskiELU()
847
+ )
848
+ (16): Sequential(
849
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
850
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
851
+ (2): MinkowskiELU()
852
+ )
853
+ (17): Sequential(
854
+ (0): MinkowskiConvolution(in=64, out=64, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
855
+ (1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
856
+ (2): MinkowskiELU()
857
+ )
858
+ )
859
+ )
860
+ (point_head): None
861
+ (roi_head): CAGroup3DRoIHead(
862
+ (proposal_target_layer): ProposalTargetLayer()
863
+ (reg_loss_func): WeightedSmoothL1Loss()
864
+ (roi_grid_pool_layers): ModuleList(
865
+ (0): SimplePoolingLayer(
866
+ (grid_conv): MinkowskiConvolution(in=64, out=128, kernel_size=[5, 5, 5], stride=[1, 1, 1], dilation=[1, 1, 1])
867
+ (grid_bn): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
868
+ (grid_relu): MinkowskiELU()
869
+ (pooling_conv): MinkowskiConvolution(in=128, out=128, kernel_size=[7, 7, 7], stride=[1, 1, 1], dilation=[1, 1, 1])
870
+ (pooling_bn): MinkowskiBatchNorm(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
871
+ )
872
+ )
873
+ (reg_fc_layers): Sequential(
874
+ (0): Linear(in_features=128, out_features=256, bias=False)
875
+ (1): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
876
+ (2): ReLU()
877
+ (3): Dropout(p=0.3, inplace=False)
878
+ (4): Linear(in_features=256, out_features=256, bias=False)
879
+ (5): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
880
+ (6): ReLU()
881
+ )
882
+ (reg_pred_layer): Linear(in_features=256, out_features=6, bias=True)
883
+ )
884
+ )
885
+ )
886
+ 2023-03-26 13:04:44,332 INFO **********************Start training scannet_models/CAGroup3D(cagroup3d-win10-scannet)**********************
887
+ 2023-03-26 17:57:27,387 INFO Epoch [ 1][ 50]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.41121925950050353, loss_bbox: 0.591667195558548, loss_cls: 0.6245615810155869, loss_sem: 0.9226177096366882, loss_vote: 0.3941664391756058, one_stage_loss: 2.944232153892517, rcnn_loss_reg: 0.09563416212797166, loss_two_stage: 0.09563416212797166,
888
+ 2023-03-27 01:05:25,842 INFO Epoch [ 1][ 100]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.6474508833885193, loss_bbox: 0.9094607937335968, loss_cls: 0.6924016952514649, loss_sem: 0.49122214019298555, loss_vote: 0.3444094204902649, one_stage_loss: 3.0849449157714846, rcnn_loss_reg: 0.8054922795295716, loss_two_stage: 0.8054922795295716,
889
+ 2023-03-27 09:09:50,953 INFO Epoch [ 1][ 150]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.6611717569828034, loss_bbox: 0.9181686878204346, loss_cls: 0.5644582629203796, loss_sem: 0.41418565332889556, loss_vote: 0.3369732141494751, one_stage_loss: 2.89495756149292, rcnn_loss_reg: 0.8431160509586334, loss_two_stage: 0.8431160509586334,
890
+ 2023-03-27 16:28:05,927 INFO Epoch [ 1][ 200]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.656439710855484, loss_bbox: 0.9153263866901398, loss_cls: 0.5242239183187485, loss_sem: 0.3945661741495132, loss_vote: 0.3303620731830597, one_stage_loss: 2.8209182739257814, rcnn_loss_reg: 0.7945010769367218, loss_two_stage: 0.7945010769367218,
891
+ 2023-03-27 23:44:41,413 INFO Epoch [ 1][ 250]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.6566858065128326, loss_bbox: 0.9171705722808838, loss_cls: 0.4788109028339386, loss_sem: 0.373558344244957, loss_vote: 0.3260516971349716, one_stage_loss: 2.7522773361206054, rcnn_loss_reg: 0.8879509460926056, loss_two_stage: 0.8879509460926056,
892
+ 2023-03-28 07:09:01,191 INFO Epoch [ 1][ 300]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.6561981821060181, loss_bbox: 0.9205144667625427, loss_cls: 0.43737293481826783, loss_sem: 0.36072677552700044, loss_vote: 0.33960326194763185, one_stage_loss: 2.7144156312942505, rcnn_loss_reg: 0.8250401616096497, loss_two_stage: 0.8250401616096497,
893
+ 2023-03-28 14:42:28,718 INFO Epoch [ 1][ 350]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.6570986831188201, loss_bbox: 0.9180864799022674, loss_cls: 0.4228892314434052, loss_sem: 0.34731213927268983, loss_vote: 0.33325146436691283, one_stage_loss: 2.6786380004882813, rcnn_loss_reg: 0.8330509012937546, loss_two_stage: 0.8330509012937546,
894
+ 2023-03-28 22:09:49,850 INFO Epoch [ 1][ 400]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.656031631231308, loss_bbox: 0.9223491895198822, loss_cls: 0.43734550893306734, loss_sem: 0.3400343120098114, loss_vote: 0.3391466856002808, one_stage_loss: 2.694907293319702, rcnn_loss_reg: 0.8316945809125901, loss_two_stage: 0.8316945809125901,
895
+ 2023-03-29 05:19:54,386 INFO Epoch [ 1][ 450]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.6560523915290832, loss_bbox: 0.9204720866680145, loss_cls: 0.40069283843040465, loss_sem: 0.32628621518611906, loss_vote: 0.32195273011922837, one_stage_loss: 2.6254562520980835, rcnn_loss_reg: 0.8233302390575409, loss_two_stage: 0.8233302390575409,
896
+ 2023-03-29 12:20:56,383 INFO Epoch [ 1][ 500]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.6548633110523224, loss_bbox: 0.9198633074760437, loss_cls: 0.37377377331256867, loss_sem: 0.3071948343515396, loss_vote: 0.31842518240213397, one_stage_loss: 2.5741204023361206, rcnn_loss_reg: 0.7916725933551788, loss_two_stage: 0.7916725933551788,
897
+ 2023-03-29 19:32:20,640 INFO Epoch [ 1][ 550]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.6493923151493073, loss_bbox: 0.9170721697807313, loss_cls: 0.3662705320119858, loss_sem: 0.3003894621133804, loss_vote: 0.3346439358592033, one_stage_loss: 2.567768402099609, rcnn_loss_reg: 0.7994219380617141, loss_two_stage: 0.7994219380617141,
898
+ 2023-03-30 02:35:54,561 INFO Epoch [ 1][ 600]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.650925452709198, loss_bbox: 0.9172565031051636, loss_cls: 0.35118326723575594, loss_sem: 0.28706444770097733, loss_vote: 0.32694552272558214, one_stage_loss: 2.533375201225281, rcnn_loss_reg: 0.7785894459486008, loss_two_stage: 0.7785894459486008,
899
+ 2023-03-30 09:21:20,835 INFO Epoch [ 1][ 650]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.6479902148246766, loss_bbox: 0.9160817635059356, loss_cls: 0.3432228803634644, loss_sem: 0.27448746263980867, loss_vote: 0.33310336887836456, one_stage_loss: 2.5148857116699217, rcnn_loss_reg: 0.818324797153473, loss_two_stage: 0.818324797153473,
900
+ 2023-03-30 16:02:35,928 INFO Epoch [ 1][ 700]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.6494181442260742, loss_bbox: 0.9183876121044159, loss_cls: 0.3300267934799194, loss_sem: 0.2648452860116959, loss_vote: 0.32385929524898527, one_stage_loss: 2.486537137031555, rcnn_loss_reg: 0.8490620160102844, loss_two_stage: 0.8490620160102844,
901
+ 2023-03-30 22:40:11,348 INFO Epoch [ 1][ 750]/[ 751] : lr: 1.000e-03, sem_thr: 0.15, loss_centerness: 0.6499875509738922, loss_bbox: 0.9197572791576385, loss_cls: 0.3292002022266388, loss_sem: 0.2616344812512398, loss_vote: 0.32180937737226484, one_stage_loss: 2.482388873100281, rcnn_loss_reg: 0.7996994721889495, loss_two_stage: 0.7996994721889495,
902
+ 2023-03-30 22:49:33,109 INFO **********************End training scannet_models/CAGroup3D(cagroup3d-win10-scannet)**********************
903
+
904
+
905
+
906
+ 2023-03-30 22:49:33,111 INFO **********************Start evaluation scannet_models/CAGroup3D(cagroup3d-win10-scannet)**********************
907
+ 2023-03-30 22:49:33,112 INFO Loading SCANNET dataset
908
+ 2023-03-30 22:49:33,161 INFO Total samples for SCANNET dataset: 312
909
+ 2023-03-30 22:49:33,168 INFO ==> Loading parameters from checkpoint C:\CITYU\CS5182\proj\CAGroup3D\output\scannet_models\CAGroup3D\cagroup3d-win10-scannet\ckpt\checkpoint_epoch_1.pth to CPU
910
+ 2023-03-30 22:49:34,337 INFO ==> Checkpoint trained from version: pcdet+0.5.2+4ae8a35+py6af8eab
911
+ 2023-03-30 22:49:34,456 INFO ==> Done (loaded 838/838)
912
+ 2023-03-30 22:49:35,263 INFO *************** EPOCH 1 EVALUATION *****************
tensorboard/events.out.tfevents.1679807081.DESKTOP-3FL13RB ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:50e140159bf7588b95f8ebc81f287987aab9b3327b6c5c1c2d03faa22e50c06e
3
+ size 506189