/
main_retrieve.py
170 lines (152 loc) · 6.3 KB
/
main_retrieve.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
import torch
import torch.nn.functional as F
from torch.utils import data
# import torch.optim as optim
import os
import sys
import argparse
import time
import csv
import numpy as np
from dataset.ucf101 import UCF101
from r2plus1d import r2plus1d_18
import moco.loader
import torchvision.transforms as transforms
import torchvision.transforms._transforms_video as transforms_video
from torchvision.datasets.samplers import RandomClipSampler, UniformClipSampler
parser = argparse.ArgumentParser()
parser.add_argument('--mode', default='finetune', type=str)
parser.add_argument('--lr', default=0.001, type=float)
parser.add_argument('--wd', default=1e-3, type=float)
parser.add_argument('--img_dim', default=112, type=int)
parser.add_argument('--test_dim', default=112, type=int)
parser.add_argument('--workers', default=8, type=int)
parser.add_argument('--channel', default=128, type=int)
parser.add_argument('--gpu', default='0,1', type=str)
parser.add_argument('--num_class', default=101, type=int)
parser.add_argument('--ft', default=10, type=float)
parser.add_argument('-cpv', '--clip_per_video', default=10, type=int, metavar='N',
help='number of frame per video clip (default: 10)')
parser.add_argument('--pretrained', default='', type=str,
help='path to moco pretrained checkpoint')
parser.add_argument('data', metavar='DIR',
help='path to dataset')
parser.add_argument('--log_dir', default='logs_moco', type=str,
help='path to the tensorboard log directory')
args = parser.parse_args()
def test(model, dataloader, mode):
model.eval()
features = []
labels = []
with torch.no_grad():
for idx, (data, _, label) in enumerate(dataloader):
data = data.to(device)
label = label.to(device)
# end = time.time()
feat = model(data.squeeze(0))
feat = feat.mean(0)
features.append(feat.detach().cpu())
labels.append(label[0].cpu())
if idx % 100 == 0:
print('already down %d in %s mode'%(idx,mode))
print(label)
# bar.finish()
features = torch.stack(features)
labels = torch.stack(labels)
torch.save(features, args.log_dir + '/%s_feat.pth.tar'%mode)
torch.save(labels, args.log_dir + '/%s_label.pth.tar'%mode)
return features, labels
def retrieve(key, query, kl, ql):
print(query.shape, key.shape, ql.shape, kl.shape)
ql = ql.reshape(-1, 1)
kl = kl.reshape(-1, 1)
# query = F.normalize(query, dim=1, p=2)
# key = F.normalize(key, dim=1, p=2)
query = query - query.mean(dim=0, keepdim=True)
key = key - key.mean(dim=0, keepdim=True)
query = F.normalize(query, dim=1, p=2)
key = F.normalize(key, dim=1, p=2)
sim = torch.matmul(query, key.transpose(0, 1))
for k in [1, 5, 10, 20, 50]:
topkval, topkidx = torch.topk(sim, k, dim=1)
acc = torch.any(kl[:, 0][topkidx]==ql, dim=1).float().mean().item()
print(acc)
def main():
global device; device = torch.device('cuda')
# Data loading code
normalize_video = transforms_video.NormalizeVideo(mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
video_augmentation = transforms.Compose(
[
transforms_video.ToTensorVideo(),
transforms_video.CenterCropVideo(args.test_dim),
normalize_video,
]
)
data_dir = os.path.join(args.data, 'data')
anno_dir = os.path.join(args.data, 'anno')
audio_augmentation = moco.loader.DummyAudioTransform()
# train_augmentation = {'video': video_augmentation_train, 'audio': audio_augmentation}
augmentation = {'video': video_augmentation, 'audio': audio_augmentation}
train_dataset = UCF101(
data_dir,
anno_dir,
16,
1,
fold=1,
train=True,
transform=augmentation,
num_workers=16
)
train_sampler = UniformClipSampler(train_dataset.video_clips, args.clip_per_video)
train_loader = torch.utils.data.DataLoader(
train_dataset, batch_size=args.clip_per_video, shuffle=False,
num_workers=args.workers, pin_memory=True, sampler=train_sampler,
multiprocessing_context="fork")
val_dataset = UCF101(
data_dir,
anno_dir,
16,
1,
fold=1,
train=False,
transform=augmentation,
num_workers=16
)
# Do not use DistributedSampler since it will destroy the testing iteration process
val_sampler = UniformClipSampler(val_dataset.video_clips, args.clip_per_video)
val_loader = torch.utils.data.DataLoader(
val_dataset, batch_size=args.clip_per_video, shuffle=False,
num_workers=args.workers, pin_memory=True, sampler=val_sampler,
multiprocessing_context="fork")
from i3d import Normalize
#model = models.__dict__[args.arch]()
model = r2plus1d_18()
model.fc = Normalize(2)
# load from pre-trained, before DistributedDataParallel constructor
if args.pretrained:
if os.path.isfile(args.pretrained):
print("=> loading checkpoint '{}'".format(args.pretrained))
checkpoint = torch.load(args.pretrained, map_location="cpu")
# rename moco pre-trained keys
state_dict = checkpoint['state_dict']
for k in list(state_dict.keys()):
# retain only encoder_q up to before the embedding layer
if k.startswith('module.encoder_q') and not k.startswith('module.encoder_q.fc'):
# remove prefix
state_dict[k[len("module.encoder_q."):]] = state_dict[k]
# delete renamed or unused k
del state_dict[k]
args.start_epoch = 0
msg = model.load_state_dict(state_dict, strict=False)
print("missing", msg.missing_keys)
# assert set(msg.missing_keys) == {"fc.weight", "fc.bias"}
print("=> loaded pre-trained model '{}'".format(args.pretrained))
else:
print("=> no checkpoint found at '{}'".format(args.pretrained))
model = model.to(device)
trainfeat, trainlabel = test(model, train_loader, 'train')
testfeat, testlabel = test(model, val_loader, 'test')
retrieve(trainfeat, testfeat, trainlabel, testlabel)
if __name__ == '__main__':
main()