-
Notifications
You must be signed in to change notification settings - Fork 1
/
torch_convert.py
171 lines (148 loc) · 6.43 KB
/
torch_convert.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
# !/usr/bin/python
# -*- coding: utf-8 -*-
# jasnei@163.com
import os
import os
from sklearn.model_selection import train_test_split
import pandas as pd
import argparse
def convert_to_img(root, train_val, save_file_name, train=True):
if(train):
f=open(root + save_file_name,'w')
data_path=root + train_val
folder_list = os.listdir(data_path)
# print(folder_list)
data = []
for index in range(len(folder_list)):
folder = folder_list[index]
# print(folder)
for root, dirs, files in os.walk(os.path.join(data_path, folder)):
# print(root)
for file in files:
image_path = root + '/'+ file # os.path.join(root, file)
data.append([image_path, index])
f.write(image_path + ' ' + str(index))
f.write('\n')
print(f'There are {len(data)} images for training.')
f.close()
else:
f=open(root+ save_file_name, 'w')
data_path=root + train_val
folder_list = os.listdir(data_path)
data = []
for index in range(len(folder_list)):
folder = folder_list[index]
for root, dirs, files in os.walk(os.path.join(data_path, folder)):
for file in files:
image_path = root + '/'+ file # os.path.join(root, file)
data.append([image_path, index])
f.write(image_path + ' ' + str(index))
f.write('\n')
print(f'There are {len(data)} images for validation.')
f.close()
def convert_label(image_csv, label, data_root, train_txt, valid_txt, train_valid_ratio):
"""
image_csv: path for the image csv
label: label path
train_test_ratio: percentage of train_test, 0.1 means 0.1 for test, 0.9 for train
"""
# load label.txt create label dict to encode the label
label_dict = dict()
with open(label, encoding='utf-8') as file:
for i, name in enumerate(file.readlines()):
label_dict[name.rstrip("\n")] = i
# print(label_dict)
# collect all the data
data_total = []
with open(image_csv, 'r', encoding='utf-8') as file:
for line in file.readlines()[1:]:
name = line.split(",")
image_path = name[0]
image_name = name[1].rstrip("\n")
data_total.append(data_root + image_path + ' ' + str(label_dict[image_name]))
# print(label_dict[image_name])
train, test = train_test_split(data_total, test_size=train_valid_ratio)
ftrain = open(train_txt, 'w')
for i in train:
ftrain.write(i)
ftrain.write('\n')
ftrain.close
ftest = open(valid_txt, 'w')
for i in test:
ftest.write(i)
ftest.write('\n')
ftest.close
print(f"There are {len(data_total)} images, {len(train)} for training, {len(test)} for validation")
def convert_split(train_dir, train_valid_ratio=0.1, random_state=0):
data_path=train_dir
txt_root = os.path.split(train_dir)[0]
txt_root_files = os.listdir(txt_root)
if "train.txt" and "valid.txt" in txt_root_files:
print("train test already split")
return
else:
folder_list = os.listdir(data_path)
# print(folder_list)
# 先判断是否是文件夹,并排序
folders = []
for index in range(len(folder_list)):
if os.path.isdir(os.path.join(data_path, folder_list[index])):
folders.append(folder_list[index])
folders.sort(key=lambda x: int(x), reverse=False)
data_total = []
for index in range(len(folders)):
folder = folders[index]
folder = os.path.join(data_path, folder)
if os.path.isdir(folder):
for root, dirs, files in os.walk(folder):
# print(root)
for file in files:
image_path = os.path.join(root, file) #root + os.sep + file
data_total.append(image_path + ' ' + str(index))
# train test split
train, test = train_test_split(data_total, test_size=train_valid_ratio, random_state=random_state)
ftrain = open(os.path.join(txt_root, "train.txt"), 'w', encoding='utf-8')
for i in train:
ftrain.write(i)
ftrain.write('\n')
ftrain.close
ftest = open(os.path.join(txt_root, "valid.txt"), 'w', encoding='utf-8')
for i in test:
ftest.write(i)
ftest.write('\n')
ftest.close
print(f"There are {len(data_total)} images, {len(train)} for training, {len(test)} for validation")
def conver_split_csv(csv_path, train_valid_ratio=0.1, random_state=0):
path_splits = csv_path.split("/")
save_path = ""
for split in path_splits[:-1]:
save_path = os.path.join(save_path, split)
save_path_files = os.listdir(save_path)
if "train.csv" and "valid.csv" in save_path_files:
print("Train Test already split up!")
return
else:
df = pd.read_csv(csv_path)
train, valid = train_test_split(df, test_size=train_valid_ratio, random_state=random_state)
train.to_csv(os.path.join(save_path, "train.csv"), index=False)
valid.to_csv(os.path.join(save_path, "valid.csv"), index=False)
print(f"Split Done! There are {len(train)} for training, {len(valid)} for validation!")
if __name__ =='__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--train_dir', type=str, default='./my_data/train_set', help='train_dir') # train data path
args = parser.parse_args()
print(args)
convert_split(args.train_dir, train_valid_ratio=0.1, random_state=1)
# label = 'label.txt'
# data_root = "./data/input/ButterflyClassification/"
# image_csv = "./data/input/ButterflyClassification/train.csv"
# train_txt = data_root + "train.txt"
# valid_txt = data_root + "valid.txt"
# convert_label(image_csv, label, data_root, train_txt, valid_txt, train_valid_ratio=0.1)
# root="./data/input/ButterflyClassificatoin/image/"
# convert_to_img('./my_data/', 'train_1/', 'train_1.txt', True)
# convert_to_img('./my_data/', 'valid_1/', 'valid_1.txt',False)
# root="./my_data/"
# convert_to_img('./my_data/', 'train/', 'train.txt', True)
# convert_to_img('./my_data/', 'valid/', 'valid.txt', False)
# conver_split_csv("My_data/Train_set/training_set.csv", train_valid_ratio=0.1)