-
Notifications
You must be signed in to change notification settings - Fork 8
/
Copy pathengine.py
248 lines (196 loc) · 10.1 KB
/
engine.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
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved
"""
Train and eval functions used in main.py
"""
import math
import os
import sys
import torch
import torch.distributed as dist
from tqdm import tqdm
from typing import Iterable
import utils.misc as utils
import utils.loss_utils as loss_utils
import utils.eval_utils as eval_utils
from utils.box_utils import xywh2xyxy
def train_one_epoch(args, model: torch.nn.Module, data_loader: Iterable,
optimizer: torch.optim.Optimizer, device: torch.device,
epoch: int, max_norm: float = 0):
model.train()
metric_logger = utils.MetricLogger(delimiter=" ")
metric_logger.add_meter('lr', utils.SmoothedValue(window_size=1, fmt='{value:.6f}'))
header = 'Epoch: [{}]'.format(epoch)
print_freq = 10
# data processing
for batch in metric_logger.log_every(data_loader, print_freq, header):
img_data, text_data, target = batch
# copy to GPU
img_data = img_data.to(device)
text_data = text_data.to(device)
target = target.to(device)
# model forward
output = model(img_data, text_data)
loss_dict = loss_utils.trans_vg_loss(output, target)
losses = sum(loss_dict[k] for k in loss_dict.keys())
# reduce losses over all GPUs for logging purposes
loss_dict_reduced = utils.reduce_dict(loss_dict)
loss_dict_reduced_unscaled = {k: v for k, v in loss_dict_reduced.items()}
losses_reduced_unscaled = sum(loss_dict_reduced_unscaled.values())
loss_value = losses_reduced_unscaled.item()
if not math.isfinite(loss_value):
print("Loss is {}, stopping training".format(loss_value))
print(loss_dict_reduced)
sys.exit(1)
optimizer.zero_grad()
losses.backward()
if max_norm > 0: # max_norm defaults to 0
torch.nn.utils.clip_grad_norm_(model.parameters(), max_norm)
optimizer.step()
metric_logger.update(loss=loss_value, **loss_dict_reduced_unscaled)
metric_logger.update(lr=optimizer.param_groups[0]["lr"])
# gather the stats from all processes
metric_logger.synchronize_between_processes()
print("Averaged stats:", metric_logger)
return {k: meter.global_avg for k, meter in metric_logger.meters.items()}
@torch.no_grad()
def validate(args, model: torch.nn.Module, data_loader: Iterable, device: torch.device):
model.eval()
metric_logger = utils.MetricLogger(delimiter=" ")
header = 'Eval:'
for batch in metric_logger.log_every(data_loader, 10, header):
img_data, text_data, target = batch
batch_size = img_data.tensors.size(0)
# copy to GPU
img_data = img_data.to(device)
text_data = text_data.to(device)
target = target.to(device)
pred_boxes = model(img_data, text_data)
miou, accu = eval_utils.trans_vg_eval_val(pred_boxes, target)
metric_logger.update_v2('miou', torch.mean(miou), batch_size)
metric_logger.update_v2('accu', accu, batch_size)
# gather the stats from all processes
metric_logger.synchronize_between_processes()
stats = {k: meter.global_avg for k, meter in metric_logger.meters.items()}
return stats
@torch.no_grad()
def evaluate(args, model: torch.nn.Module, data_loader: Iterable, device: torch.device):
model.eval()
pred_box_list = []
gt_box_list = []
for _, batch in enumerate(tqdm(data_loader)):
img_data, text_data, target = batch
batch_size = img_data.tensors.size(0)
# copy to GPU
img_data = img_data.to(device)
text_data = text_data.to(device)
target = target.to(device)
output = model(img_data, text_data)
pred_box_list.append(output.cpu())
gt_box_list.append(target.cpu())
pred_boxes = torch.cat(pred_box_list, dim=0)
gt_boxes = torch.cat(gt_box_list, dim=0)
total_num = gt_boxes.shape[0]
accu_num = eval_utils.trans_vg_eval_test(pred_boxes, gt_boxes)
result_tensor = torch.tensor([accu_num, total_num]).to(device)
torch.cuda.synchronize()
dist.all_reduce(result_tensor)
accuracy = float(result_tensor[0]) / float(result_tensor[1])
return accuracy
@torch.no_grad() # this is iou-based
def evaluate_for_filtering(args, model: torch.nn.Module, data_loader: Iterable, device: torch.device):
# TODO: ATTENTION: evaluate_for_filtering() only can running on the single GPU env, to prevent the data missing.
# besides, evaluate_for_filtering must use collate_fn=utils.collate_fn_filtering
model.eval()
pred_box_list = []
gt_box_list = []
img_file_list = []
phrase_list = []
img_iou_list = []
bbox_ori_list = []
train_pseudo_bad_label_0_0 = []
train_pseudo_bad_label_0_4 = []
train_pseudo_refine = []
reliability = [i / 10.0 for i in range(11)] # default delta is 0.1, the reliability interval is from 0.0 to 1.0
for i in reliability:
train_pseudo_refine.append([])
for _, batch in enumerate(tqdm(data_loader)):
img_data, text_data, target, img_file, phrase, bbox_ori = batch
# copy to GPU
img_data = img_data.to(device)
text_data = text_data.to(device)
target = target.to(device)
output = model(img_data, text_data)
pred_box_list.append(output.cpu())
gt_box_list.append(target.cpu())
for img_name in img_file:
img_file_list.append(img_name)
for phrase_i in phrase:
phrase_list.append(phrase_i)
for bbox_ori_i in bbox_ori:
bbox_ori_list.append(bbox_ori_i)
pred_boxes = torch.cat(pred_box_list, dim=0)
gt_boxes = torch.cat(gt_box_list, dim=0)
total_num = gt_boxes.shape[0]
accu_num, iou_list = eval_utils.trans_vg_eval_test_iou(pred_boxes, gt_boxes)
result_tensor = torch.tensor([accu_num, total_num]).to(device)
for i in range(len(img_file_list)):
if not (args.dataset == 'flickr'): # unc/+/g， bbox xywh, referit:x1y1
img_iou_list.append([img_file_list[i], 'useless placeholder', bbox_ori_list[i], phrase_list[i],
gt_boxes[i].cpu(), pred_boxes[i].cpu(), float(iou_list[i])])
if float(iou_list[i]) == 0:
tmp_pseudo_label = [img_file_list[i], 'useless placeholder', bbox_ori_list[i], phrase_list[i],
'useless placeholder', float(iou_list[i])]
train_pseudo_bad_label_0_0.append(tmp_pseudo_label)
if float(iou_list[i]) < 0.4:
tmp_pseudo_label = [img_file_list[i], 'useless placeholder', bbox_ori_list[i], phrase_list[i],
'useless placeholder', float(iou_list[i])]
train_pseudo_bad_label_0_4.append(tmp_pseudo_label)
for j in range(len(reliability) - 1):
if float(iou_list[i]) > reliability[j]:
tmp_pseudo_label = [img_file_list[i], 'useless placeholder', bbox_ori_list[i], phrase_list[i],
'useless placeholder']
train_pseudo_refine[j].append(tmp_pseudo_label)
if float(iou_list[i]) == 1.0:
tmp_pseudo_label = [img_file_list[i], 'useless placeholder', bbox_ori_list[i], phrase_list[i],
'useless placeholder']
train_pseudo_refine[10].append(tmp_pseudo_label)
else:
if args.dataset == 'flickr':
# tmp_sample = [sample[0], sample[2], sample[3]], ['img_file', [x1y1x2y2 bbox], 'expression']
img_iou_list.append([img_file_list[i], 'useless placeholder', bbox_ori_list[i], phrase_list[i],
gt_boxes[i].cpu(), pred_boxes[i].cpu(), float(iou_list[i])])
if float(iou_list[i]) == 0:
tmp_pseudo_label = [img_file_list[i], 'useless placeholder', bbox_ori_list[i], phrase_list[i],
'useless placeholder', float(iou_list[i])]
train_pseudo_bad_label_0_0.append(tmp_pseudo_label)
if float(iou_list[i]) < 0.4:
tmp_pseudo_label = [img_file_list[i], 'useless placeholder', bbox_ori_list[i], phrase_list[i],
'useless placeholder', float(iou_list[i])]
train_pseudo_bad_label_0_4.append(tmp_pseudo_label)
for j in range(len(reliability) - 1):
if float(iou_list[i]) > reliability[j]:
tmp_pseudo_label = [img_file_list[i], bbox_ori_list[i], phrase_list[i]]
train_pseudo_refine[j].append(tmp_pseudo_label)
if float(iou_list[i]) == 1.0:
tmp_pseudo_label = [img_file_list[i], bbox_ori_list[i], phrase_list[i]]
train_pseudo_refine[10].append(tmp_pseudo_label)
print("img_file_list: ", len(img_file_list))
print("iou_list: ", iou_list.size())
print("img_iou_list size: ", len(img_iou_list))
print("train_pseudo_bad_label_0_0 size: ", len(train_pseudo_bad_label_0_0))
print("train_pseudo_bad_label_0_4 size: ", len(train_pseudo_bad_label_0_4))
for j in range(len(reliability)):
print("the number of train_pseudo_refine sample when reliability > ", str(reliability[j]), " is: ",
len(train_pseudo_refine[j]))
torch.save(train_pseudo_bad_label_0_0, os.path.join(args.output_dir, '{}_train_pseudo_bad_label_0_0.pth'.format(args.dataset)))
torch.save(train_pseudo_bad_label_0_4, os.path.join(args.output_dir, '{}_train_pseudo_bad_label_0_4.pth'.format(args.dataset)))
for j in range(len(reliability)):
torch.save(train_pseudo_refine[j], os.path.join(args.output_dir,
'{}_train_pseudo_refine_{}.pth'.format(args.dataset, str(reliability[j]).replace(".", "_"))))
torch.save(img_iou_list, os.path.join(args.output_dir, '{}_train_pseudo_iou_result.pth'.format(args.dataset)))
torch.cuda.synchronize()
dist.all_reduce(result_tensor)
all_img_iou_list = []
accuracy = float(result_tensor[0]) / float(result_tensor[1])
print("img_iou_list: ", len(all_img_iou_list))
return accuracy, all_img_iou_list