-
Notifications
You must be signed in to change notification settings - Fork 10
/
train_HDPrior_pretrain.py
185 lines (158 loc) · 7.71 KB
/
train_HDPrior_pretrain.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
import os
import numpy as np
import sys
import logging
import time
from tensorboardX import SummaryWriter
import argparse
import matplotlib.pyplot as plt
import torch
import torch.nn as nn
from tools.config import Config
from torch.optim.lr_scheduler import StepLR
from tools.loss import SimpleLoss, DiscriminativeLoss
from data_osm.dataset import semantic_dataset
from data_osm.const import NUM_CLASSES
from tools.evaluation.iou import get_batch_iou
from tools.evaluation.angle_diff import calc_angle_diff
from tools.eval import onehot_encoding, eval_pretrain
from model.utils.map_mae_head import vit_base_patch8
from model import get_model
import warnings
warnings.filterwarnings("ignore")
from collections import OrderedDict
def write_log(writer, ious, title, counter):
writer.add_scalar(f'{title}/iou', torch.mean(ious[1:]), counter)
for i, iou in enumerate(ious):
writer.add_scalar(f'{title}/class_{i}/iou', iou, counter)
def train(cfg):
if not os.path.exists(cfg.logdir):
os.makedirs(cfg.logdir)
logging.basicConfig(filename=os.path.join(cfg.logdir, "results.log"),
filemode='w',
format='%(asctime)s: %(message)s',
datefmt='%Y-%m-%d %H:%M:%S',
level=logging.INFO)
logging.getLogger('shapely.geos').setLevel(logging.CRITICAL)
logger = logging.getLogger()
logger.addHandler(logging.StreamHandler(sys.stdout))
data_conf = {
'num_channels': NUM_CLASSES + 1,
'image_size': cfg.image_size,
'xbound': cfg.xbound,
'ybound': cfg.ybound,
'zbound': cfg.zbound,
'dbound': cfg.dbound,
'thickness': cfg.thickness,
'angle_class': cfg.angle_class,
'patch_w': cfg.patch_w,
'patch_h': cfg.patch_h,
'mask_ratio': cfg.mask_ratio,
'mask_flag': cfg.mask_flag,
'sd_map_path': cfg.sd_map_path,
}
train_loader, val_loader = semantic_dataset(cfg, cfg.version, cfg.dataroot, data_conf,
cfg.batch_size, cfg.nworkers, cfg.dataset)
patch_h = data_conf['ybound'][1] - data_conf['ybound'][0]
patch_w = data_conf['xbound'][1] - data_conf['xbound'][0]
canvas_h = int(patch_h / data_conf['ybound'][2])
canvas_w = int(patch_w / data_conf['xbound'][2])
# # TODO: add to cfg and add support for patch32
# model = vit_base_patch8(data_conf=data_conf,
# instance_seg=cfg.instance_seg,
# embedded_dim=cfg.embedding_dim,
# direction_pred=cfg.direction_pred,
# direction_dim=cfg.angle_class,
# lidar=True,
# img_size=(canvas_h, canvas_w))
model = get_model(cfg, data_conf, cfg.instance_seg, cfg.embedding_dim, cfg.direction_pred, cfg.angle_class)
if 'vit_base' in cfg and cfg.vit_base is not None:
state_dict_model = torch.load(cfg.vit_base)
model.load_state_dict(state_dict_model, strict=False)
model = nn.DataParallel(model, device_ids=cfg.gpus)
opt = torch.optim.Adam(model.parameters(), lr=cfg.lr, weight_decay=cfg.weight_decay)
sched = StepLR(opt, 3, 0.1)
writer = SummaryWriter(logdir=cfg.logdir)
loss_fn = SimpleLoss(cfg.pos_weight).cuda()
embedded_loss_fn = DiscriminativeLoss(cfg.embedding_dim, cfg.delta_v, cfg.delta_d).cuda()
direction_loss_fn = torch.nn.BCELoss(reduction='none')
model.cuda(device=cfg.gpus[0])
model.train()
counter = 0
last_idx = len(train_loader) - 1
for epoch in range(cfg.nepochs):
for batchi, (imgs, trans, rots, intrins, post_trans, post_rots,
lidar_data, lidar_mask, car_trans, yaw_pitch_roll,
semantic_gt, instance_gt, direction_gt, osm_masks,
osm_vectors, masked_map, timestamps, scene_ids) in enumerate(train_loader):
t0 = time.time()
opt.zero_grad()
semantic, embedding, direction = model(masked_map.float())
semantic_gt = semantic_gt.cuda().float()
instance_gt = instance_gt.cuda()
device = semantic_gt.device
if semantic.device != device:
semantic = semantic.to(device)
embedding = embedding.to(device)
direction = direction.to(device)
seg_loss = loss_fn(semantic, semantic_gt)
if cfg.instance_seg:
var_loss, dist_loss, reg_loss = embedded_loss_fn(embedding, instance_gt)
else:
var_loss = 0
dist_loss = 0
reg_loss = 0
if cfg.direction_pred:
direction_gt = direction_gt.cuda()
lane_mask = (1 - direction_gt[:, 0]).unsqueeze(1)
direction_loss = direction_loss_fn(torch.softmax(direction, 1), direction_gt)
direction_loss = (direction_loss * lane_mask).sum() / (lane_mask.sum() * direction_loss.shape[1] + 1e-6)
angle_diff = calc_angle_diff(direction, direction_gt, cfg.angle_class)
else:
direction_loss = 0
angle_diff = 0
final_loss = seg_loss * cfg.scale_seg + var_loss * cfg.scale_var + dist_loss * cfg.scale_dist + direction_loss * cfg.scale_direction
final_loss.backward()
torch.nn.utils.clip_grad_norm_(model.parameters(), cfg.max_grad_norm)
opt.step()
counter += 1
t1 = time.time()
if counter % 100 == 0:
intersects, union = get_batch_iou(onehot_encoding(semantic), semantic_gt)
iou = intersects / (union + 1e-7)
logger.info(f"TRAIN[{epoch:>3d}]: [{batchi:>4d}/{last_idx}] "
f"Time: {t1-t0:>7.4f} "
f"Loss: {final_loss.item():>7.4f} "
f"IOU: {np.array2string(iou[1:].numpy(), precision=3, floatmode='fixed')}")
write_log(writer, iou, 'train', counter)
writer.add_scalar('train/step_time', t1 - t0, counter)
writer.add_scalar('train/seg_loss', seg_loss, counter)
writer.add_scalar('train/var_loss', var_loss, counter)
writer.add_scalar('train/dist_loss', dist_loss, counter)
writer.add_scalar('train/reg_loss', reg_loss, counter)
writer.add_scalar('train/direction_loss', direction_loss, counter)
writer.add_scalar('train/final_loss', final_loss, counter)
writer.add_scalar('train/angle_diff', angle_diff, counter)
cur_lr = opt.state_dict()['param_groups'][0]['lr']
writer.add_scalar('train/lr', cur_lr, counter)
model_name = os.path.join(cfg.logdir, f"model{epoch}.pt")
torch.save(model.state_dict(), model_name)
logger.info(f"{model_name} saved")
iou = eval_pretrain(model, val_loader)
logger.info(f"EVAL[{epoch:>2d}]: "
f"IOU: {np.array2string(iou[1:].numpy(), precision=3, floatmode='fixed')}")
write_log(writer, iou, 'eval', counter)
model.train()
sched.step()
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='P-MapNet pre-train HD Prior.')
parser.add_argument("config", help = 'path to config file', type=str, default=None)
args = parser.parse_args()
cfg = Config.fromfile(args.config)
if not os.path.exists(cfg.logdir):
os.makedirs(cfg.logdir)
with open(os.path.join(cfg.logdir, 'config.txt'), 'w') as f:
argsDict = cfg.__dict__
for eachArg, value in argsDict.items():
f.writelines(eachArg + " : " + str(value) + "\n")
train(cfg)