forked from Rodger-Huang/SYSU-HCP-at-ImageCLEF-VQA-Med-2021
-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathinference_ensemble_once.py
174 lines (150 loc) · 6.8 KB
/
inference_ensemble_once.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
import argparse
import json
import os
from tqdm import tqdm
import torch
import torch.nn as nn
from torch.utils.data import DataLoader
from torchvision import transforms
import csv
from dataloaders import custom_transforms as trforms
from dataloaders import tnsc_dataset
from utils import flip_lr
from torchvision.models.resnet import resnet34, resnet18, resnet50, resnet101
from torchvision.models.vgg import vgg16, vgg16_bn, vgg19, vgg19_bn
from torchvision.models.densenet import densenet169, densenet121, densenet201
from resnest.torch import resnest50, resnest101
from model.vgg_hgap import vgg16HGap, vgg16HGap_bn, vgg19HGap, vgg16_multi_scales, \
vgg16_add, vgg16HGap_resizer
from model.resnet_hgap import resnet50HGap
from model.resnest_hgap import resnest50HGap
from model.repvgg import get_RepVGG_func_by_name
from model import pretrainedmodels
def get_arguments():
parser = argparse.ArgumentParser()
parser.add_argument('-gpu', type=str, default='0')
parser.add_argument('-backbone', type=str, default='resnest50')
parser.add_argument('-input_size', type=int, default=224)
parser.add_argument('-model_path', type=str, default='path2model')
parser.add_argument('-classes', type=int, default=330)
return parser.parse_args()
modal_mask = json.load(open('modelmask.json', 'r'))
modalties = modal_mask.keys()
def main(args, test_pbar):
backbone, input_size, model_path = args
os.environ["CUDA_VISIBLE_DEVICES"] = '6'
if backbone == 'resnet18' or backbone == 'resnet34' or \
backbone == 'resnet50' or backbone == 'resnet101' or \
backbone == 'resnest50' or backbone == 'resnest101':
backbone = eval(backbone)(pretrained=True)
backbone.fc = nn.Linear(in_features=backbone.fc.in_features, out_features=330)
elif backbone == 'vgg16' or backbone == 'vgg16_bn' or \
backbone == 'vgg19' or backbone == 'vgg19_bn':
backbone = eval(backbone)(pretrained=True)
backbone.classifier = nn.Sequential(
nn.Linear(512 * 7 * 7, 4096),
nn.ReLU(True),
nn.Dropout(),
nn.Linear(4096, 4096),
nn.ReLU(True),
nn.Dropout(),
nn.Linear(4096, 330),
)
elif backbone == 'vgg16HGap' or backbone == 'vgg16HGap_bn' or \
backbone == 'vgg19HGap' or backbone == 'vgg16HGap_resizer' or \
backbone == 'vgg16_multi_scales' or \
backbone == 'vgg16_add' or \
backbone == 'resnet50HGap' or \
backbone == 'resnest50HGap':
backbone = eval(backbone)(pretrained=True, num_classes=330)
elif backbone == 'densenet121' or backbone == 'densenet169' or \
backbone == 'densenet201':
backbone = eval(backbone)(pretrained=True)
backbone.classifier = nn.Linear(in_features=backbone.classifier.in_features, out_features=330)
elif backbone == 'RepVGG-B0' or backbone == 'RepVGG-B1' or backbone == 'RepVGG-B2':
backbone = get_RepVGG_func_by_name(backbone)(deploy=False)
backbone.load_state_dict(torch.load('checkpoint/RepVGG/{}-train.pth'.format(backbone)))
backbone.linear = nn.Linear(in_features=backbone.linear.in_features, out_features=330)
elif backbone == 'inceptionresnetv2' or backbone == 'inceptionv4' or \
backbone == 'xception' or backbone == 'polynet' or \
backbone == 'se_resnet50':
backbone = eval('pretrainedmodels.' + backbone)(pretrained='imagenet')
backbone.last_linear = nn.Linear(in_features=backbone.last_linear.in_features, out_features=330, bias=True)
else:
raise NotImplementedError
model = torch.load(model_path, map_location='cpu')
state_dict = backbone.state_dict()
for k in state_dict:
k_m = k
if not model['state_dict'].__contains__(k_m):
k_m = 'module.' + k
state_dict[k] = model['state_dict'][k_m]
backbone.load_state_dict(state_dict)
torch.cuda.set_device(device=0)
backbone.cuda()
backbone.eval()
pred_list = []
for sample_batched in test_pbar:
q = sample_batched['question'][0]
mask = []
for modality in modalties:
if modality in q:
mask = modal_mask[modality]
img = sample_batched['image'].cuda()
feats = backbone(img)
prob, pred_idx = torch.topk(feats, 20, dim=1)
if prob[0][0].item() > 0.5:
pred = pred_idx[0][0]
pred_list.append(pred.item())
continue
if len(mask) != 0:
for i in range(len(pred_idx[0])):
if pred_idx[0][i] in mask:
pred = i
break
if i == 7:
pred = pred_idx[0][0]
else:
pred = pred_idx[0][0]
# print(pred)
# img_flip = flip_lr(img)
# feats_flip = backbone(img_flip)
# feats_pp = (feats + feats_flip) / 2
# pred_pp = torch.argmax(feats_pp, dim=1, keepdim=False)
pred_list.append(pred.item())
return pred_list
if __name__ == '__main__':
arg_list = [
('resnet50', 256, 'VQA-MED-2021-Models/res50/models/backbone_epoch-49_1.0000_1.0000.pth'),
('resnet50HGap', 256, 'VQA-MED-2021-Models/res50hgap/models/backbone_epoch-49_1.0000_1.0000.pth'),
('resnest50', 256, 'VQA-MED-2021-Models/ress50/models/backbone_epoch-29_0.9980_0.9980.pth'),
('resnest50HGap', 256, 'VQA-MED-2021-Models/ress50hgap/models/backbone_epoch-29_1.0000_1.0000.pth'),
('vgg16', 224, 'VQA-MED-2021-Models/vgg16/models/backbone_epoch-49_1.0000_1.0000.pth'),
('vgg16HGap', 224, 'VQA-MED-2021-Models/vgg16hgap/models/backbone_epoch-49_1.0000_1.0000.pth'),
('vgg19', 224, 'VQA-MED-2021-Models/vgg19/models/backbone_epoch-49_1.0000_1.0000.pth'),
('vgg19HGap', 224, 'VQA-MED-2021-Models/vgg19hgap/models/backbone_epoch-49_1.0000_1.0000.pth'),
]
composed_transforms_test = transforms.Compose([
trforms.FixedResizeI(size=(384, 384)),
trforms.NormalizeI(mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225)),
trforms.ToTensorI()])
testset = tnsc_dataset.MedLTDataset(mode='test', path='test2021', transform=composed_transforms_test, return_size=False)
testloader = DataLoader(testset, batch_size=1, shuffle=False, num_workers=2)
test_pbar = tqdm(testloader, unit_scale=1)
result = []
for arg in arg_list:
preds = main(arg, test_pbar)
result.append(preds)
print(result)
new_result = []
for i in range(500):
votes = []
for res in result:
votes.append(res[i])
new_result.append(votes)
ensemble_result = []
for i in range(500):
ensemble_result.append(max(new_result[i], key=new_result[i].count))
with open("./res.txt", "a", newline='') as f:
for i in ensemble_result:
f.write(str(i)+'\n')