forked from hujan01/DenseWAP
-
Notifications
You must be signed in to change notification settings - Fork 0
/
test_classic.py
195 lines (158 loc) · 5.65 KB
/
test_classic.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
'''
Author: sigmoid
Description: 基于传统注意力模块的测试脚本
Email: 595495856@qq.com
Date: 2021-03-06 19:38:18
LastEditTime: 2021-03-06 20:10:26
'''
import math
import os
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
import cv2
from PIL import Image
import numpy as np
import torch
import torch.nn as nn
from torch.autograd import Variable
import torch.nn.functional as F
import torch.utils.data as data
from dataset import MERData
from model_classic import Encoder, Decoder
from config import cfg
from utils.util import get_all_dist, load_dict, custom_dset, collate_fn_double
torch.backends.cudnn.benchmark = False
# 配置参数
valid_datasets = ['data/valid.pkl', 'data/CROHME2016/label/test_caption_2014.txt']
dictionaries = 'data/CROHME2016/label/dictionary.txt'
result_path = "results/recognition.txt"
Imagesize = 500000
batch_size_t = 2
maxlen = 70
maxImagesize = 100000
hidden_size = 256
worddicts = load_dict(dictionaries) #token 2 id
worddicts_r = [None] * len(worddicts) #id 2 token
for kk, vv in worddicts.items():
worddicts_r[vv] = kk
# 数据读取处理
test, test_label, uidList = MERData(
valid_datasets[0], valid_datasets[1], worddicts, batch_size=1,
batch_Imagesize=Imagesize, maxlen=maxlen, maxImagesize=maxImagesize)
image_test = custom_dset(test, test_label)
test_loader = torch.utils.data.DataLoader(
dataset = image_test,
batch_size = batch_size_t,
shuffle = True,
collate_fn = collate_fn_double,
num_workers = cfg.num_workers,
)
# 1. 加载模型
encoder = Encoder(img_channels=2)
decoder = Decoder(112)
encoder = encoder.cuda()
decoder = decoder.cuda()
encoder.load_state_dict(torch.load('checkpoints/encoder_classic.pkl'))
decoder.load_state_dict(torch.load('checkpoints/attn_decoder_classic.pkl'))
encoder.eval()
decoder.eval()
# 评估参数
total_dist = 0 # 统计所有的序列的总编辑距离
total_label = 0 # 统计所有序列的总长度
total_line = 0 # 统计一共有多少个序列
total_line_rec = 0 # 统计识别正确的序列
error1, error2, error3 = 0, 0, 0
fw = open(result_path, 'w') # 保存识别结果
# 2. 开始评估
for step_t, (x_t, y_t, batch_list) in enumerate(test_loader):
# abandon <batch data
if x_t.size()[0]<batch_size_t:
break
x_t = x_t.cuda()
y_t = y_t.cuda()
feat_t = encoder(x_t) # (bs, c, h, w) c=684
# 1.init input
decoder_input_t = torch.LongTensor([111]*batch_size_t).view(-1, 1).cuda()
decoder_hidden_t = decoder.init_hidden(batch_size_t).cuda()
# 2.reset coverage
decoder.reset(batch_size_t, feat_t.size())
prediction = torch.zeros(batch_size_t, maxlen)
prediction_sub = []
prediction_real = []
label_sub = []
label_real = []
# 处理标签
m = torch.nn.ZeroPad2d((0, maxlen-y_t.size()[1], 0, 0))
y_t = m(y_t)
for i in range(maxlen):
decoder_output_t, decoder_hidden_t, _ = decoder(decoder_input_t, decoder_hidden_t, feat_t)
topv, topi = torch.max(decoder_output_t, 1)
if torch.sum(topi)==0: # 一个批次中所有序列都预测完成
break
decoder_input_t = topi
decoder_input_t = decoder_input_t.view(batch_size_t, 1)
# prediction
prediction[:, i] = decoder_input_t.flatten()
for i in range(batch_size_t):
uid = uidList[batch_list[i]]
for j in range(maxlen):
if int(prediction[i][j]) == 0:
break
else:
prediction_sub.append(int(prediction[i][j]))
prediction_real.append(worddicts_r[int(prediction[i][j])])
if len(prediction_sub) < maxlen: #不足后面填0
prediction_sub.append(0)
for k in range(y_t.size()[1]):
if int(y_t[i][k]) == 0:
break
else:
label_sub.append(int(y_t[i][k]))
label_real.append(worddicts_r[int(y_t[i][k])])
label_sub.append(0)
# 评价指标
dist, llen, sub, ins, dls = get_all_dist(label_sub, prediction_sub)
wer_step = float(dist) / llen
if len(prediction_sub)==len(label_sub) : # 计算error1, error2
e = 0
for r, l in zip(prediction_sub, label_sub):
if r!=l :
e += 1
if e>3: break # 超过3个直接跳出
if e==1:
error1 += 1
elif e==2:
error2 += 1
elif e==3:
error3 += 1
wer_step = float(dist) / llen
total_dist += dist
total_label += llen
total_line += 1
if dist == 0:
total_line_rec = total_line_rec + 1
print('step is %d' % (step_t))
print('prediction is ')
print(prediction_real)
print('the truth is')
print(label_real)
print('the wer is %.5f' % (wer_step))
# save predict result
fw.write(uid+'\t')
fw.write(' '.join(prediction_real)+'\n')
label_sub = []
prediction_sub = []
label_real = []
prediction_real = []
fw.close()
wer = float(total_dist) / total_label
print("{}/{}".format(total_line_rec, total_line))
exprate = float(total_line_rec) / total_line
error1 += total_line_rec
error2 += error1
error3 += error2
# 打印评测结果
print('{}/{}'.format(total_line_rec, total_line))
print('ExpRate is {:.4f}'.format(exprate))
print('error1 nums: {}, error2 nums: {}, error3 nums: {}'.format(error1, error2, error3))
print('error1 is {:.4f}, error2 is {:.4f}, error3 is {:.4f}'.format((error1)/total_line, error2/total_line, error3/total_line))
print('wer is {:.4f}'.format(wer))