-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathphase2.py
186 lines (148 loc) · 6.6 KB
/
phase2.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
import os
import sys
import numpy as np
import torch
from torch import nn, optim
from torch.optim.lr_scheduler import ReduceLROnPlateau
from torch.utils.data import DataLoader
from tqdm import tqdm
# our code
import pi
from datasets import davis17
from models.unet_model import UNetMedium, UNetSmall
sys.path.append(os.path.dirname(os.path.dirname(os.path.realpath(__file__))))
from utils import utils
CHECKPOINTS_FOLDER = "checkpoints_sftrackpp"
def forward_batch(num_trackers, models, sfseg_params, batch, batch_idx,
device):
rgb_frames, gt_segm_imgs, trackers_bbox_imgs = batch
bs, num_frames, chan, h, w = rgb_frames.shape
M0 = sfseg_params["M0"]
rgb_frames = rgb_frames.to(device=device)
gt_segm_imgs = gt_segm_imgs.to(device=device).float()
trackers_bbox_imgs = trackers_bbox_imgs.to(device=device).float()
# rgb_frames: BS x 2*M0 + 1 x channels x H x W
# trackers_bbox_imgs: x num_trackers x
# gt_segm_imgs: x 1 x
# Phase1. Output from all trackers
phase1_segms = []
for tr_idx in range(num_trackers):
# concatenate on channels axis
tracker_inp = torch.cat(
[rgb_frames, trackers_bbox_imgs[:, :, tr_idx:tr_idx + 1]], axis=2)
segm_pred1 = models[0](tracker_inp.view(bs * num_frames, 4, h,
w)).view(
bs, num_frames, 1, h, w)
phase1_segms.append(segm_pred1)
# Phase 2. SFSeg
input_masks = torch.cat(phase1_segms, axis=2)
phase2_segm_interm = utils.sfsegpp(models[1],
input_masks=input_masks,
trackers_output=input_masks,
sfseg_params=sfseg_params)[:, M0:M0 + 1]
# phase2_segm_interm shape: BS x 1 x H x W
phase2_segm = models[2](phase2_segm_interm)
return phase2_segm, gt_segm_imgs
def train_phase2(epoch, trackers, models, data_loader, sfseg_params, optimizer,
scheduler, loss_fcn, device):
for model in models:
model.train()
training_loss = 0
num_trackers = len(trackers)
M0 = sfseg_params["M0"]
for batch_idx, batch in enumerate(tqdm(data_loader)):
phase2_segm, gt_segm_imgs = forward_batch(num_trackers, models,
sfseg_params, batch,
batch_idx, device)
batch_loss = loss_fcn(phase2_segm, gt_segm_imgs[:, M0])
crt_batch_loss = batch_loss.item()
training_loss += crt_batch_loss
batch_loss.backward()
# Optimizer
optimizer.step()
optimizer.zero_grad()
scheduler.step(crt_batch_loss)
training_loss /= len(data_loader)
return training_loss
def val_phase2(epoch, trackers, models, data_loader, sfseg_params, loss_fcn,
device):
for model in models:
model.eval()
val_loss = 0
num_trackers = len(trackers)
M0 = sfseg_params["M0"]
with torch.no_grad():
for batch_idx, batch in enumerate(tqdm(data_loader)):
phase2_segm, gt_segm_imgs = forward_batch(num_trackers, models,
sfseg_params, batch,
batch_idx, device)
batch_loss = loss_fcn(phase2_segm, gt_segm_imgs[:, M0])
crt_batch_loss = batch_loss.item()
val_loss += crt_batch_loss
val_loss /= len(data_loader)
return val_loss
def main():
n_epochs = 2
device = "cuda" if torch.cuda.is_available() else "cpu"
trackers = ["dimp", "atom", "segm", "siamban", "siamrpnpp"]
trackers = ["dimp"]
kernel_size = (3, 5, 5)
M0 = kernel_size[0] // 2
ds_train = davis17.Davis17AllTrackersDataset(trackers,
"train",
M0=M0,
samples_per_video=2)
ds_val = davis17.Davis17AllTrackersDataset(trackers,
"val",
M0=M0,
samples_per_video=2)
dl_train = DataLoader(ds_train, batch_size=7, shuffle=True, num_workers=20)
dl_val = DataLoader(ds_val, batch_size=30, shuffle=True, num_workers=20)
# models
net_phase1 = UNetMedium(n_inp=4, n_outp=1, with_dropout=False)
utils.load_model(net_phase1,
"%s/phase1_net1_basic.pth" % CHECKPOINTS_FOLDER)
net_phase1.to(device)
net_phase1 = nn.DataParallel(net_phase1)
net_phase2 = nn.Conv2d(in_channels=len(trackers),
out_channels=1,
kernel_size=1,
bias=True)
net_phase2.to(device)
net_phase2_1 = UNetSmall(n_inp=1, n_outp=1, with_dropout=False)
net_phase2_1.to(device)
net_phase2_1 = nn.DataParallel(net_phase2_1)
models = [net_phase1, net_phase2, net_phase2_1]
all_params = utils.chain_generators(net_phase1.parameters(),
net_phase2.parameters(),
net_phase2_1.parameters())
optimizer = optim.SGD(all_params,
lr=0.02,
weight_decay=1e-4,
nesterov=True,
momentum=0.9)
scheduler = ReduceLROnPlateau(optimizer,
patience=10,
factor=0.1,
threshold=0.005,
min_lr=1e-4,
verbose=True)
loss_fcn = nn.BCEWithLogitsLoss(pos_weight=torch.tensor(5.)).to(device)
sfseg_params = {}
M0 = kernel_size[0] // 2
sfseg_params["M0"] = M0
sfseg_params["filter"] = pi.init_cnn_filter(kernel_size)
sfseg_params["p"] = 0.1
sfseg_params["alpha"] = 0.5
for epoch in range(n_epochs):
train_phase2(epoch, trackers, models, dl_train, sfseg_params,
optimizer, scheduler, loss_fcn, device)
val_phase2(epoch, trackers, models, dl_val, sfseg_params, loss_fcn,
device)
utils.save_model(net_phase1.module,
"%s/phase2_net1.pth" % CHECKPOINTS_FOLDER)
utils.save_model(net_phase2, "%s/phase2_net2.pth" % CHECKPOINTS_FOLDER)
utils.save_model(net_phase2_1.module,
"%s/phase2_net2_1.pth" % CHECKPOINTS_FOLDER)
if __name__ == '__main__':
main()