-
Notifications
You must be signed in to change notification settings - Fork 11
/
train_scpnet_comp.py
220 lines (174 loc) · 9.38 KB
/
train_scpnet_comp.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
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
# -*- coding:utf-8 -*-
# author: Xinge, Xzy
# @file: train_cylinder_asym.py
import os
import time
import argparse
import sys
import numpy as np
import torch
import torch.optim as optim
from tqdm import tqdm
# from utils.metric_util import per_class_iu, fast_hist_crop
from dataloader.pc_dataset import get_SemKITTI_label_name, get_eval_mask, unpack
from builder import data_builder, model_builder, loss_builder
from config.config import load_config_data
from utils.load_save_util import load_checkpoint
import warnings
from utils.np_ioueval import iouEval
import yaml
warnings.filterwarnings("ignore")
def main(args):
pytorch_device = torch.device('cuda:0')
config_path = args.config_path
configs = load_config_data(config_path)
dataset_config = configs['dataset_params']
train_dataloader_config = configs['train_data_loader']
val_dataloader_config = configs['val_data_loader']
val_batch_size = val_dataloader_config['batch_size']
train_batch_size = train_dataloader_config['batch_size']
model_config = configs['model_params']
train_hypers = configs['train_params']
grid_size = model_config['output_shape']
num_class = model_config['num_class']
ignore_label = dataset_config['ignore_label']
model_load_path = train_hypers['model_load_path']
model_save_path = train_hypers['model_save_path']
SemKITTI_label_name = get_SemKITTI_label_name(dataset_config["label_mapping"])
unique_label = np.asarray(sorted(list(SemKITTI_label_name.keys())))[1:] - 1
unique_label_str = [SemKITTI_label_name[x] for x in unique_label + 1]
my_model = model_builder.build(model_config)
model_load_path += '0.pth'
model_save_path += ''
if os.path.exists(model_load_path):
print('Load model from: %s' % model_load_path)
my_model = load_checkpoint(model_load_path, my_model)
else:
print('No existing model, training model from scratch...')
if not os.path.exists(model_save_path):
os.makedirs(model_save_path)
print(model_save_path)
my_model.to(pytorch_device)
optimizer = optim.Adam(my_model.parameters(), lr=train_hypers["learning_rate"])
loss_func, lovasz_softmax = loss_builder.build(wce=True, lovasz=True,
num_class=num_class, ignore_label=ignore_label)
train_dataset_loader, val_dataset_loader, val_pt_dataset = data_builder.build(dataset_config,
train_dataloader_config,
val_dataloader_config,
grid_size=grid_size,
use_tta=False,
use_multiscan=True)
# training
epoch = 0
best_val_miou = 0
my_model.train()
global_iter = 0
check_iter = train_hypers['eval_every_n_steps']
# learning map
with open("config/label_mapping/semantic-kitti.yaml", 'r') as stream:
semkittiyaml = yaml.safe_load(stream)
class_strings = semkittiyaml["labels"]
class_inv_remap = semkittiyaml["learning_map_inv"]
while epoch < train_hypers['max_num_epochs']:
loss_list = []
pbar = tqdm(total=len(train_dataset_loader))
time.sleep(10)
# lr_scheduler.step(epoch)
for i_iter, (_, train_vox_label, train_grid, _, train_pt_fea, train_index, origin_len) in enumerate(train_dataset_loader):
if global_iter % check_iter == 0 and epoch > 0:
my_model.eval()
val_loss_list = []
val_method = 2 # 1-segmentation method, 2-completion method
if val_method == 1:
hist_list = []
else:
evaluator = iouEval(num_class, [])
with torch.no_grad():
for i_iter_val, (_, val_vox_label, val_grid, _, val_pt_fea, val_index, origin_len) in enumerate(
val_dataset_loader):
val_pt_fea_ten = [torch.from_numpy(i).type(torch.FloatTensor).to(pytorch_device) for i in
val_pt_fea]
val_grid_ten = [torch.from_numpy(i).to(pytorch_device) for i in val_grid]
for bat in range(val_batch_size):
val_label_tensor = val_vox_label[bat,:].type(torch.LongTensor).to(pytorch_device)
val_label_tensor = torch.unsqueeze(val_label_tensor, 0)
predict_labels = my_model(val_pt_fea_ten, val_grid_ten, val_batch_size)
loss = lovasz_softmax(torch.nn.functional.softmax(predict_labels).detach(), val_label_tensor,
ignore=ignore_label) + loss_func(predict_labels.detach(), val_label_tensor)
predict_labels = torch.argmax(predict_labels, dim=1)
predict_labels = predict_labels.cpu().detach().numpy()
predict_labels = np.squeeze(predict_labels)
val_vox_label0 = val_vox_label[bat, :].cpu().detach().numpy()
val_vox_label0 = np.squeeze(val_vox_label0)
val_name = val_pt_dataset.im_idx[val_index[0]]
invalid_name = val_name.replace('velodyne', 'voxels')[:-3]+'invalid'
invalid_voxels = unpack(np.fromfile(invalid_name, dtype=np.uint8)) # voxel labels
invalid_voxels = invalid_voxels.reshape((256, 256, 32))
masks = get_eval_mask(val_vox_label0, invalid_voxels)
predict_labels = predict_labels[masks]
val_vox_label0 = val_vox_label0[masks]
evaluator.addBatch(predict_labels.astype(int), val_vox_label0.astype(int))
val_loss_list.append(loss.detach().cpu().numpy())
# my_model.train()
print('Validation per class iou: ')
_, class_jaccard = evaluator.getIoU()
m_jaccard = class_jaccard[1:].mean()
iou = class_jaccard
val_miou = m_jaccard * 100
ignore = [0]
# print also classwise
for i, jacc in enumerate(class_jaccard):
if i not in ignore:
print('IoU class {i:} [{class_str:}] = {jacc:.3f}'.format(
i=i, class_str=class_strings[class_inv_remap[i]], jacc=jacc))
# compute remaining metrics.
conf = evaluator.get_confusion()
acc_completion = (np.sum(conf[1:, 1:])) / (np.sum(conf) - conf[0, 0])
print('Current val completion iou is %.3f' % acc_completion)
del val_vox_label, val_grid, val_pt_fea, val_pt_fea_ten, val_grid_ten, val_label_tensor
# save model if performance is improved
if best_val_miou < val_miou:
best_val_miou = val_miou
# save model with best val miou for completion
model_save_name = model_save_path + ('iou%.4f_epoch%d.pth' % (val_miou, epoch))
torch.save(my_model.state_dict(), model_save_name)
print('Current val miou is %.3f while the best val miou is %.3f' %
(val_miou, best_val_miou))
print('Current val loss is %.3f' %
(np.mean(val_loss_list)))
my_model.train()
train_pt_fea_ten = [torch.from_numpy(i).type(torch.FloatTensor).to(pytorch_device) for i in train_pt_fea]
train_vox_ten = [torch.from_numpy(i).to(pytorch_device) for i in train_grid]
point_label_tensor = train_vox_label.type(torch.LongTensor).to(pytorch_device)
# forward + backward + optimize
outputs = my_model(train_pt_fea_ten, train_vox_ten, point_label_tensor.shape[0])
loss = lovasz_softmax(torch.nn.functional.softmax(outputs), point_label_tensor, ignore=ignore_label) + loss_func(
outputs, point_label_tensor)
loss.backward()
optimizer.step()
loss_list.append(loss.item())
if global_iter % 1000 == 0:
if len(loss_list) > 0:
print('epoch %d iter %5d, loss: %.3f\n' %
(epoch, i_iter, np.mean(loss_list)))
else:
print('loss error')
optimizer.zero_grad()
pbar.update(1)
global_iter += 1
if global_iter % check_iter == 0:
if len(loss_list) > 0:
print('epoch %d iter %5d, loss: %.3f\n' %
(epoch, i_iter, np.mean(loss_list)))
else:
print('loss error')
pbar.close()
epoch += 1
if __name__ == '__main__':
# Training settings
parser = argparse.ArgumentParser(description='')
parser.add_argument('-y', '--config_path', default='config/semantickitti-multiscan.yaml')
args = parser.parse_args()
print(' '.join(sys.argv))
print(args)
main(args)