Document_Classifier/GPUmodelLoad1.py at master · HaKaTaHan/Document_Classifier · GitHub

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
#-*- coding:utf-8 -*-
import os
import torch
import torch.nn as nn
import torch.nn.parallel
import torch.utils.data
import torchvision.datasets as dset
import torchvision.transforms as transforms
import numpy as np
import random
from collections import OrderedDict

torch.manual_seed(0)
np.random.seed(0)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
torch.backends.cudnn.enabled = False
random.seed(999)
torch.manual_seed(999)

class Generator(nn.Module):
    def __init__(self, ngpu):
        self.__ngf = 32
        self.__ndf = 32
        self.__nc = 3
        super(Generator, self).__init__()
        self.__ngpu = ngpu
        self.__main = nn.Sequential(
            nn.ConvTranspose2d(self.__ngf * 4, self.__ngf * 4, 4, 1, 0, bias=False),
            nn.BatchNorm2d(self.__ngf * 4),
            nn.ReLU(True),
            nn.ConvTranspose2d(self.__ngf * 4, self.__ngf * 2, 4, 2, 1, bias=False),
            nn.BatchNorm2d(self.__ngf * 2),
            nn.ReLU(True),
            nn.ConvTranspose2d(self.__ngf * 2, self.__ngf, 4, 2, 1, bias=False),
            nn.BatchNorm2d(self.__ngf),
            nn.ReLU(True),
            nn.ConvTranspose2d(self.__ngf, self.__nc, 4, 2, 1, bias=False),
            nn.Tanh()
        )

    def forward(self, input):
        return self.__main(input)

class Discriminator(nn.Module):
    def __init__(self, ngpu):
        self.__ngf = 32
        self.__ndf = 32
        self.__nc = 3
        super(Discriminator, self).__init__()
        self.__ngpu = ngpu
        self.__main = nn.Sequential(
            nn.Conv2d(self.__nc, self.__ndf, 4, 2, 1, bias=False),
            nn.LeakyReLU(0.2, inplace=True),
            nn.Conv2d(self.__ndf, self.__ndf * 2, 4, 2, 1, bias=False),
            nn.BatchNorm2d(self.__ndf * 2),
            nn.LeakyReLU(0.2, inplace=True),
            nn.Conv2d(self.__ndf * 2, self.__ndf * 4, 4, 2, 1, bias=False),
            nn.BatchNorm2d(self.__ndf * 4),
            nn.LeakyReLU(0.2, inplace=True),
            nn.Conv2d(self.__ndf * 4, 1, 4, 1, 0, bias=False),
            nn.Sigmoid()
        )

    def forward(self, input):
        return self.__main(input)

class model(object):
    # def __init__(self, ocr, ip, crop_path, improvement_path):
    def __init__(self):
        self.__modelD = Discriminator(1)
        self.__dataroot = './IMG'
        self.__img_size = 32
        self.__batch_size = 1
        self.__workers = 2
        self.__train_data_name = os.listdir(self.__dataroot+'/IMG')
        self.__train_data = dset.ImageFolder(root = self.__dataroot,
                                             transform = transforms.Compose([
                                                 transforms.Resize(self.__img_size),
                                                 transforms.CenterCrop(self.__img_size),
                                                 transforms.ToTensor(),
                                                 transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)), ]))
        self.__dataloader = torch.utils.data.DataLoader(self.__train_data,
                                                        batch_size = self.__batch_size,
                                                        num_workers = self.__workers)
        self.__ngpu = 1
        self.__device = torch.device("cuda:0" if (
                torch.cuda.is_available() and self.__ngpu > 0) else "cpu")
        self.__ngf = 32
        self.__ndf = 32
        self.__nc = 3
        self.__criterion = nn.BCELoss()

        self.__curList = []
        self.__coverList = []
        self.__dict_errD = {}

    def showGan(self):
        print("showGan", self.__device)

        self.__modelD = nn.DataParallel(self.__modelD)

        checkpointD = torch.load('./modelData/ver_999_dModel_100_11.pt', map_location=torch.device('cpu'))
        state_dict_D = checkpointD['model_state_dict']
        new_state_dict_D = OrderedDict()

        for k, v in state_dict_D.items():
            k = 'module._Discriminator__'+k
            new_state_dict_D[k] = v

        self.__modelD.load_state_dict(new_state_dict_D)
        self.__modelD.eval()
        predict_list = []

        for i, data in enumerate(self.__dataloader, 0):

            real_cpu = data[0].to(self.__device)
            b_size = real_cpu.size(0)
            label = torch.full((b_size,), 1, device = self.__device)

            output = self.__modelD(real_cpu).view(-1)

            errD_real = self.__criterion(output, label)

            errD = errD_real
            self.__dict_errD[self.__train_data_name[i]] = errD.item()
            print(self.__train_data_name[i], errD)
            if errD < 2.3:
                predict_list.append(self.__train_data_name[i])

        self.__coverList = predict_list

    def identify_cover(self):
        return self.__coverList

    def dictErrD(self):
        return self.__dict_errD


if __name__ == '__main__':
    pass