forked from ErlerPhilipp/points2surf
-
Notifications
You must be signed in to change notification settings - Fork 0
/
full_run.py
151 lines (131 loc) · 5.81 KB
/
full_run.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
import os
from source import points_to_surf_train
from source import points_to_surf_eval
from source import sdf
from source.base import evaluation
# When you see this error:
# 'Expected more than 1 value per channel when training...' which is raised by the BatchNorm1d layer
# for multi-gpu, use a batch size that can't be divided by the number of GPUs
# for single-gpu, use a straight batch size
# see https://github.com/pytorch/pytorch/issues/2584
# see https://forums.fast.ai/t/understanding-code-error-expected-more-than-1-value-per-channel-when-training/9257/12
if __name__ == '__main__':
model_name = 'vanilla'
dataset = 'abc_minimal'
base_dir = 'datasets'
in_dir_train = os.path.join(base_dir, dataset)
train_set = 'trainset.txt'
val_set = 'valset.txt'
test_set = 'testset.txt'
# features = ['imp_surf', 'patch_pts_ids', 'p_index'] # l2-loss
features = ['imp_surf_magnitude', 'imp_surf_sign', 'patch_pts_ids', 'p_index'] # l2-loss + BCE-loss
# workers = 22 # for strong training machine
workers = 7 # for typical PC
# batch_size = 501 # ~7.5 GB memory on 4 2080 TI for 300 patch points + 1000 sub-sample points
# batch_size = 3001 # ~10 GB memory on 4 2080 TI for 50 patch points + 200 sub-sample points
batch_size = 100 # ~7 GB memory on 1 1070 for 300 patch points + 1000 sub-sample points
# grid_resolution = 256 # quality like in the paper
grid_resolution = 128 # quality for a short test
rec_epsilon = 3
certainty_threshold = 13
sigma = 5
fixed_radius = False
patch_radius = 0.1 if fixed_radius else 0.0
single_transformer = 0
shared_transformer = 0
# os.environ["CUDA_VISIBLE_DEVICES"] = "0"
train_params = [
'--name', model_name,
'--desc', model_name,
'--indir', in_dir_train,
'--outdir', 'models',
'--trainset', train_set,
'--testset', val_set,
'--net_size', str(1024),
'--nepoch', str(10),
'--lr', str(0.01),
'--debug', str(0),
'--workers', str(workers),
'--batchSize', str(batch_size),
'--points_per_patch', str(300),
'--patches_per_shape', str(1000),
'--sub_sample_size', str(1000),
'--cache_capacity', str(10),
'--patch_radius', str(patch_radius),
'--single_transformer', str(single_transformer),
'--shared_transformer', str(shared_transformer),
'--patch_center', 'mean',
'--training_order', 'random_shape_consecutive',
'--use_point_stn', str(1),
'--uniform_subsample', str(0),
'--outputs',
]
train_params += features
# train model on GT data with multiple query points per patch
train_opt = points_to_surf_train.parse_arguments(train_params)
points_to_surf_train.points_to_surf_train(train_opt)
valsets = ['abc_minimal', ]
for valset in valsets:
# validate model
in_dir_val = os.path.join(base_dir, valset)
out_dir_val = os.path.join('results', model_name, valset)
res_dir_eval = os.path.join(out_dir_val, 'eval')
eval_params = [
'--indir', in_dir_val,
'--outdir', out_dir_val,
'--dataset', val_set,
'--models', model_name,
'--batchSize', str(batch_size),
'--workers', str(workers),
'--cache_capacity', str(5),
]
eval_opt = points_to_surf_eval.parse_arguments(eval_params)
points_to_surf_eval.points_to_surf_eval(eval_opt)
evaluation.eval_predictions(
os.path.join(res_dir_eval, 'eval'),
os.path.join(in_dir_val, '05_query_dist'),
os.path.join(res_dir_eval, 'rme_comp_res.csv'),
unsigned=False)
testsets = ['abc_minimal', ]
for testset in testsets:
out_dir = os.path.join('results', model_name, testset)
res_dir_rec = os.path.join(out_dir, 'rec')
# reconstruct SDFs from testset
in_dir_test = os.path.join(base_dir, testset)
print('Points2Surf is reconstructing {} into {}'.format(out_dir, res_dir_rec))
recon_params = [
'--indir', in_dir_test,
'--outdir', out_dir,
'--dataset', test_set,
'--query_grid_resolution', str(grid_resolution),
'--reconstruction', str(True),
'--models', model_name,
'--batchSize', str(batch_size),
'--workers', str(workers),
'--cache_capacity', str(5),
'--epsilon', str(rec_epsilon),
]
recon_opt = points_to_surf_eval.parse_arguments(recon_params)
points_to_surf_eval.points_to_surf_eval(recon_opt)
# reconstruct meshes from predicted SDFs
imp_surf_dist_ms_dir = os.path.join(res_dir_rec, 'dist_ms')
query_pts_ms_dir = os.path.join(res_dir_rec, 'query_pts_ms')
vol_out_dir = os.path.join(res_dir_rec, 'vol')
mesh_out_dir = os.path.join(res_dir_rec, 'mesh')
sdf.implicit_surface_to_mesh_directory(
imp_surf_dist_ms_dir, query_pts_ms_dir,
vol_out_dir, mesh_out_dir,
grid_resolution, sigma, certainty_threshold,
workers)
# get Hausdorff distance for reconstructed meshes
new_meshes_dir_abs = os.path.join(res_dir_rec, 'mesh')
ref_meshes_dir_abs = os.path.join(in_dir_test, '03_meshes')
csv_file = os.path.join(res_dir_rec, 'hausdorff_dist_pred_rec.csv')
evaluation.mesh_comparison(
new_meshes_dir_abs=new_meshes_dir_abs,
ref_meshes_dir_abs=ref_meshes_dir_abs,
num_processes=workers,
report_name=csv_file,
samples_per_model=10000,
dataset_file_abs=os.path.join(in_dir_test, test_set))
print('Points2Surf is finished!')