-
Notifications
You must be signed in to change notification settings - Fork 1
/
test.py
54 lines (47 loc) · 1.88 KB
/
test.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
from dataloader.cub_dataset import CUB
from model.bn_inception import bn_inception
import numpy as np
import random
import torch
import torchvision.transforms as trsfrm
from dataloader.trsfrms import must_transform
from evaluation.recall import give_recall
from torch.utils.data import DataLoader
import argparse
import logging
import os
if __name__ == '__main__':
# seeds
seed = 1000
random.seed(seed)
np.random.seed(seed)
torch.manual_seed(seed)
torch.cuda.manual_seed_all(seed)
parser = argparse.ArgumentParser(description='MVR')
parser.add_argument("--exp_name", default="exp", type=str)
parser.add_argument("--model_save_dir", default="./MVR_Triplet/exp", type=str)
args = parser.parse_args()
log_dir = os.path.join("./log", args.exp_name)
if os.path.exists(log_dir) != True:
os.makedirs(log_dir)
logging.basicConfig(
format="%(asctime)s %(message)s",
level=logging.INFO,
handlers=[
logging.FileHandler("{0}/{1}/test.log".format('log', args.exp_name)),
logging.StreamHandler()
]
)
cuda = 'cuda:0'
root_dir = "./data/CUB_200_2011/images"
transforms_test = trsfrm.Compose([must_transform(), trsfrm.Resize(256), trsfrm.CenterCrop(224)])
cub_test = CUB(root_dir, 'test', transforms_test)
test_loader = DataLoader(cub_test, batch_size=64, num_workers=8, shuffle=False, pin_memory=True)
saved_model_dir = args.model_save_dir + "/best.pth"
net = bn_inception(embedding_size=64, pretrained=True, is_norm=True, bn_freeze=False)
net.load_state_dict(torch.load(saved_model_dir))
net.to(cuda)
net.eval()
with torch.no_grad():
Recalls = give_recall(net, test_loader,cuda= cuda, visualize=True, dataset=cub_test)
logging.info("Recall@1 = {}, Recall@2 = {}, Recall@4 = {}, Recall@8 = {}".format(Recalls[0], Recalls[1],Recalls[2], Recalls[3]))