def load_file_csv(path, file_name):
path_list = extract_all_path(path, 'csv')
data = pd.read_csv(path_list[0])
for path_file in path_list[1:]:
data1 = pd.read_csv(path_file)
data = data.append(data1)
data.to_csv(path + file_name + '.csv', index=False)
EDA_data_label(path + file_name + '.csv', path + file_name + '.png')
def split_train_test_data(path):
path_list = extract_all_path(path, 'csv')
for path_file in path_list:
data = pd.read_csv(path_file)
data_group = data.groupby(['names_image', 'list_label'],
axis=0).count().reset_index()
data_group = data_group.sample(len(data_group))
msk = np.random.rand(len(data_group)) < 0.8
train_merge = data_group[msk]
test_merge = data_group[~msk]
print(train_merge.shape)
print(test_merge.shape)
# train.to_csv('data/data_0930_merge_2/train_merge.csv',index=False)
# test.to_csv('data/data_0930_merge_2/test_merge.csv', index=False)
# data = pd.read_csv('data/data_0930_merge_2/data_0930_merge_2.csv')
# train_label = pd.read_csv('data/data_0930_merge_2/train_merge.csv')
train = data[~data['names_image'].isin(train_merge['names_image'])]
test = data[~data['names_image'].isin(test_merge['names_image'])]
if len(train) > len(test):
train.to_csv('data/data_0930_merge_2/data_unmerge/train/' +
path_file.split('/')[-1],
index=False)
EDA_data_label(
'data/data_0930_merge_2/data_unmerge/train/' +
path_file.split('/')[-1],
'data/data_0930_merge_2/data_unmerge/train/' +
(path_file.split('/')[-1]).split('.')[0] + '.png')
test.to_csv('data/data_0930_merge_2/data_unmerge/test/' +
path_file.split('/')[-1],
index=False)
EDA_data_label(
'data/data_0930_merge_2/data_unmerge/test/' +
path_file.split('/')[-1],
'data/data_0930_merge_2/data_unmerge/test/' +
(path_file.split('/')[-1]).split('.')[0] + '.png')
else:
train.to_csv('data/data_0930_merge_2/data_unmerge/test/' +
path_file.split('/')[-1],
index=False)
EDA_data_label(
'data/data_0930_merge_2/data_unmerge/test/' +
path_file.split('/')[-1],
'data/data_0930_merge_2/data_unmerge/test/' +
(path_file.split('/')[-1]).split('.')[0] + '.png')
test.to_csv('data/data_0930_merge_2/data_unmerge/train/' +
path_file.split('/')[-1],
index=False)
EDA_data_label(
'data/data_0930_merge_2/data_unmerge/train/' +
path_file.split('/')[-1],
'data/data_0930_merge_2/data_unmerge/train/' +
(path_file.split('/')[-1]).split('.')[0] + '.png')
for img in image_list:
img = img.replace("unmerge", "merge_1")
yield(folder + img + '.png')
# data_words = list(sent_to_words(df_train['Review']))
# ten anh tuong ung
def label(path_list, folder, path_file_csv, file_name):
my_array2 = np.genfromtxt(path_list, dtype=str,
skip_header=0)
image_list = list(create_path_image(folder, my_array2))
df_info_image = pd.DataFrame()
try:
df_info_image = extract_info_image(image_list)
df_info_image.to_csv(path_file_csv)
EDA_data_label(path_file_csv, file_name)
except Exception:
print(Exception)
return df_info_image
path = 'data_0930_merge_1_tmp/Segmentation/'
folder = 'data_0930_merge_1_tmp/masks/'
path_file_text = extract_all_path(path, 'txt')
for path_list in path_file_text:
save_path = 'data/data_0930_merge_1/' + (str(path_list).split('/')[-1]).split('.')[0]
path_file_csv = save_path + '.csv'
file_name = save_path + '.png'
df_info_inmage = label(path_list, folder, path_file_csv, file_name)
print(df_info_inmage)
import os
import pandas as pd
from extract_information_image import extract_all_path
path = []
path_file = extract_all_path('crawl_data/pngimg.com1/', 'png')
list_str = ['_', '.']
# list_str = ['_PNG_', '_Transparent', '_Background', '_Download']
# for pt in path_file:
# path.append((pt.split('/'))[-1].split('_')[0])
for pt in path_file:
for x in list_str:
if x in pt:
pa = (pt.split('/')[-1]).split(x)[0]
pt = pa
path.append(pa)
df = pd.DataFrame()
df['Path'] = path_file
df['Category'] = path
df.to_csv('crawl_data/path_file_pngimg.csv', index=False)