-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathtrain_cpn.py
193 lines (141 loc) · 8 KB
/
train_cpn.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
import numpy as np
import torch
from torch.backends import cudnn
cudnn.enabled = True
from torch.utils.data import DataLoader
from torchvision import transforms
import voc12.data
from tool import pyutils, imutils, torchutils,visualization
import argparse
import importlib
import torch.nn.functional as F
import os
torch.backends.cudnn.benchmark = True
def adaptive_min_pooling_loss(x):
# This loss does not affect the highest performance, but change the optimial background score (alpha)
n,c,h,w = x.size()
k = h*w//4
x = torch.max(x, dim=1)[0]
y = torch.topk(x.view(n,-1), k=k, dim=-1, largest=False)[0]
y = F.relu(y, inplace=False)
loss = torch.sum(y)/(k*n)
return loss
if __name__ == '__main__':
session_name = 'CPN'
results_path = 'results/' + session_name
if not os.path.exists(results_path):
os.makedirs(results_path)
parser = argparse.ArgumentParser()
parser.add_argument("--batch_size", default=4, type=int)
parser.add_argument("--max_epoches", default=8, type=int)
parser.add_argument("--network", default="network.resnet38_CPN", type=str)
parser.add_argument("--lr", default=0.01, type=float)
parser.add_argument("--num_workers", default=2, type=int)
parser.add_argument("--wt_dec", default=5e-4, type=float)
parser.add_argument("--weights", required = True , type=str) #path for the ilsvrc-cls_rna-a1_cls1000_ep-0001.pth
parser.add_argument("--train_list", default="voc12/train_aug.txt", type=str)
parser.add_argument("--val_list", default="voc12/val.txt", type=str)
parser.add_argument("--session_name", default=session_name, type=str)
parser.add_argument("--crop_size", default=448, type=int)
parser.add_argument("--voc12_root", required=True, type=str) #path for the VOC12, .../VOCdevkit/VOC2012
args = parser.parse_args()
model = getattr(importlib.import_module(args.network), 'Net')()
pyutils.Logger(results_path+'/'+args.session_name + '.log')
print(vars(args))
train_dataset = voc12.data.VOC12ClsDataset(args.train_list, voc12_root=args.voc12_root,
transform=transforms.Compose([
imutils.RandomResizeLong(448, 768),
transforms.RandomHorizontalFlip(),
transforms.ColorJitter(brightness=0.3, contrast=0.3, saturation=0.3, hue=0.1),
np.asarray,
model.normalize,
imutils.RandomCrop(args.crop_size),
imutils.HWC_to_CHW,
torch.from_numpy,
]))
train_data_loader = DataLoader(train_dataset, batch_size=args.batch_size,
shuffle=True, num_workers=args.num_workers, pin_memory=False, drop_last=True)
max_step = (len(train_dataset) // args.batch_size) * args.max_epoches
param_groups = model.get_parameter_groups()
optimizer = torchutils.PolyOptimizer([
{'params': param_groups[0], 'lr': args.lr, 'weight_decay': args.wt_dec},
{'params': param_groups[1], 'lr': 2*args.lr, 'weight_decay': 0},
{'params': param_groups[2], 'lr': 10*args.lr, 'weight_decay': args.wt_dec},
{'params': param_groups[3], 'lr': 20*args.lr, 'weight_decay': 0}
], lr=args.lr, weight_decay=args.wt_dec, max_step=max_step)
if args.weights[-7:] == '.params':
assert args.network == "network.resnet38_cls"
import network.resnet38d
weights_dict = network.resnet38d.convert_mxnet_to_torch(args.weights)
else:
weights_dict = torch.load(args.weights)
model.load_state_dict(weights_dict, strict=False)
model.cuda()
model.train()
avg_meter1 = pyutils.AverageMeter('loss', 'loss_cls', 'loss_tcp', 'loss_cpcr')
timer = pyutils.Timer("Session started: ")
for ep in range(args.max_epoches):
for iter, pack in enumerate(train_data_loader):
img = pack[1].cuda()
N, C, H, W = img.size()
label = pack[2]
bg_score = torch.ones((N,1))
label = torch.cat((bg_score, label), dim=1)
label = label.cuda().unsqueeze(2).unsqueeze(3)
img1 = pack[3].cuda()
img2 = pack[4].cuda()
ctri = pack[5].view(N, 1, 1, 1).repeat(1, 21, 1, 1).type(torch.FloatTensor).cuda()
uctri = pack[6].view(N, 1, 1, 1).repeat(1, 21, 1, 1).type(torch.FloatTensor).cuda()
cam,cam_improved = model(img)
l = F.adaptive_avg_pool2d(cam, (1, 1))
loss_rvmin = adaptive_min_pooling_loss((cam_improved * label)[:, 1:, :, :])
cam = visualization.max_norm(cam)*label
cam_improved = visualization.max_norm(cam_improved)*label
cam1,cam_improved1 = model(img1)
l1 = F.adaptive_avg_pool2d(cam1, (1, 1))
loss_rvmin1 = adaptive_min_pooling_loss((cam_improved1 * label)[:, 1:, :, :])
cam1 = ctri*visualization.max_norm(cam1)*label
cam_improved1 = ctri*visualization.max_norm(cam_improved1)*label
cam2,cam_improved2 = model(img2)
l2 = F.adaptive_avg_pool2d(cam2,(1,1))
loss_rvmin2 = adaptive_min_pooling_loss((cam_improved2 * label)[:, 1:, :, :])
cam2= uctri*visualization.max_norm(cam2) * label
cam_improved2 = uctri*visualization.max_norm(cam_improved2) * label
loss_cls = F.multilabel_soft_margin_loss(l[:,1:,:,:],label[:,1:,:,:])
loss_cls1 = F.multilabel_soft_margin_loss(l1[:,1:,:,:],label[:,1:,:,:])
loss_cls2 = F.multilabel_soft_margin_loss(l2[:,1:,:,:],label[:,1:,:,:])
loss_cls = (loss_cls+loss_cls1+loss_cls2) / 3 + (loss_rvmin+loss_rvmin1+loss_rvmin2) / 3
loss_tcp = torch.mean(torch.abs((cam2[:,1:,:,:].detach() + cam1[:,1:,:,:].detach()) - cam[:,1:,:,:]))
loss_tcp_improved = torch.mean(torch.abs(cam_improved[:,1:,:,:]-cam_improved1[:,1:,:,:]-cam_improved2[:,1:,:,:]))
loss_tcp_all = loss_tcp+loss_tcp_improved
ns, cs, hs, ws = cam.size()
cam[:, 0, :, :] = 1 - torch.max(cam[:, 1:, :, :], dim=1)[0]
cam1[:, 0, :, :] = 1 - torch.max(cam1[:, 1:, :, :], dim=1)[0]
cam2[:, 0, :, :] = 1 - torch.max(cam2[:, 1:, :, :], dim=1)[0]
cam_improved[:, 0, :, :] = 1 - torch.max(cam_improved[:, 1:, :, :], dim=1)[0]
cam_improved1[:, 0, :, :] = 1 - torch.max(cam_improved1[:, 1:, :, :], dim=1)[0]
cam_improved2[:, 0, :, :] = 1 - torch.max(cam_improved2[:, 1:, :, :], dim=1)[0]
loss_cpcr1 = torch.abs(imutils.max_onehot(cam.detach()-cam1.detach())-cam_improved2)
loss_cpcr2 = torch.abs(imutils.max_onehot(cam.detach()-cam2.detach())-cam_improved1)
loss_cpcr1_ohem = torch.mean(torch.topk(loss_cpcr1.view(ns,-1), k=(int)(21*hs*ws*0.2), dim=-1)[0]) #OHEM is 20%
loss_cpcr2_ohem = torch.mean(torch.topk(loss_cpcr2.view(ns,-1), k=(int)(21*hs*ws*0.2), dim=-1)[0])
#
loss_cpcr = loss_cpcr1_ohem + loss_cpcr2_ohem
loss = loss_cls+loss_tcp_all+loss_cpcr
avg_meter1.add({'loss': loss.item(),'loss_cls':loss_cls.item(),'loss_tcp':loss_tcp_all.item(),
'loss_cpcr':loss_cpcr.item()
})
optimizer.zero_grad()
loss.backward()
optimizer.step()
if (optimizer.global_step-1)%1 == 0:
timer.update_progress(optimizer.global_step / max_step)
print('Iter:%5d/%5d' % (optimizer.global_step - 1, max_step),
'loss:%.4f %.4f %.4f %.4f' % avg_meter1.get('loss', 'loss_cls', 'loss_tcp', 'loss_cpcr'),
'imps:%.1f' % ((iter+1) * args.batch_size / timer.get_stage_elapsed()),
'Fin:%s' % (timer.str_est_finish()),
'lr: %.4f' % (optimizer.param_groups[0]['lr']), flush=True)
avg_meter1.pop()
else:
timer.reset_stage()
torch.save(model.state_dict(), results_path+'/'+args.session_name +'.pth')