def gen_11(tmp_dir):
img_paths = get_all_file_from_dir(tmp_dir)
alph, alph_label = get_num_path(img_paths, 4)
six, six_label = get_num_path(img_paths, 6)
one, one_label = get_num_path(img_paths, 1)
if alph is not None and six is not None:
alph = cv2.imread(alph)
six = cv2.imread(six)
one = cv2.imread(one)
alph_shape = alph.shape
six_shape = six.shape
one_shape = one.shape
if alph_shape[0] / alph_shape[1] > 1 and six_shape[0] / six_shape[
1] > 1: # vertical
alph = cv2.resize(alph,
(32, int(32 / alph_shape[1] * alph_shape[0])))
six = cv2.resize(six, (32, int(32 / six_shape[1] * six_shape[0])))
one = cv2.resize(one, (32, int(32 / one_shape[1] * one_shape[0])))
return np.concatenate((alph, six, one),
axis=0), alph_label + six_label + one_label
if alph_shape[0] / alph_shape[1] < 1 and six_shape[0] / six_shape[
1] < 1: # horizontal
alph = cv2.resize(alph,
(int(32 / alph_shape[0] * alph_shape[1]), 32))
six = cv2.resize(six, (int(32 / six_shape[0] * six_shape[1]), 32))
one = cv2.resize(one, (int(32 / one_shape[0] * one_shape[1]), 32))
return np.concatenate((alph, six, one),
axis=1), alph_label + six_label + one_label
return None, None
def load_all_data():
files_paths = get_all_file_from_dir(IMAGE_PATH)
print(('training images size:' + str(len(files_paths))))
print('start load training data')
data = {'one': {}, 'four': {}, 'six': {}, 'zero': []}
for p in files_paths:
img = cv2.imread(p)
shape = img.shape
img = img.reshape((1, shape[0], shape[1], shape[2]))
_, image_name = os.path.split(p)
label = image_name.replace('#', '').split('-')[0]
if len(label) == 1:
if label not in data['one'].keys():
data['one'][label] = []
data['one'][label].append({'label': label, 'img': img})
elif len(label) == 4:
if label not in data['four'].keys():
data['four'][label] = []
data['four'][label].append({'label': label, 'img': img})
elif len(label) == 6:
if label not in data['six'].keys():
data['six'][label] = []
data['six'][label].append({'label': label, 'img': img})
else:
data['zero'].append({'label': label, 'img': img})
print('loading data finish')
return data
def load_map(path):
paths = get_all_file_from_dir(path)
map = {}
for p in paths:
dir_, name = os.path.split(p)
key = name.split('.')[0]
map[key] = p
return map
def get_all_orig_char_path():
"""
获取所有初始图片文本
:return:
"""
dir_path = GENERATE_DIE_PATH + 'orig'
file_paths = get_all_file_from_dir(dir_path)
return file_paths
def get_unrelated_image():
"""
get image path that do not relate annotation path
:return:
"""
unrelated_image = []
an_path = get_all_file_from_dir(ANNOTATION_PATH)
image_path = get_all_file_from_dir(IMAGE_PATH)
for p in image_path:
is_related = False
name = p.split('\\')[-1].split('.')[0]
for a in an_path:
if a.find(name) > 0:
is_related = True
if not is_related:
unrelated_image.append(p)
return unrelated_image
def main():
all_image_path = get_all_file_from_dir(HORIZONTAL_PATH)
index = 0
for path in all_image_path:
index += 1
if index % 1000 == 0:
print(index)
new_img = get_resize_image(path)
new_img = np.rot90(new_img)
save_img(new_img, path)
def load_annotation_info():
"""
load annotation info
:return:
"""
an_path = get_all_file_from_dir(ANNOTATION_PATH)
annotation_info = []
for path in an_path:
parse_info = parse_xml_file(path)
annotation_info.append(parse_info)
return annotation_info
def load_image_info():
"""
load image info
:return:
"""
image_paths = get_all_file_from_dir(IMAGE_PATH)
image_map = {}
for p in image_paths:
name = p.split('\\')[-1]
image_map[name] = p
return image_paths
def main():
all_image_path = get_all_file_from_dir(HORIZONTAL_PATH)
length = len(all_image_path)
index = 0
for path in all_image_path:
index += 1
if index % 100 == 0:
print(str(index * 1.0 / length))
new_img = get_resize_image(path)
# print(path)
save_img(new_img, path)
def load_image_info(dir):
"""
load image info
:return:
"""
image_paths = get_all_file_from_dir(dir)
image_map = {}
for p in image_paths:
name = p.split('/')[-1].split('.')[0]
image_map[name] = p
return image_map
def main():
dir_path = 'E:\dataset/text_area\horizontal\horizontal_orig/'
paths = get_all_file_from_dir(dir_path)
for i, p in enumerate(paths):
if i % 100 == 0:
print(i)
_, img_name = os.path.split(p)
label = img_name.split('-')[0]
if len(label) != 6:
continue
img = cv2.imread(p)
seg_horozontal(label, img)
def main():
index = 0
paths = get_all_file_from_dir(ori_path)
for index, p in enumerate(paths):
seg = int(index / 2000) + 1
_, name = os.path.split(p)
label = name.split('-')[0]
new_name = str(index) + '-' + str(uuid.uuid4()) + '.jpg'
index += 1
d_path = des_path + new_name
shutil.copy(p, d_path)
def main():
paths = get_all_file_from_dir(JSON_DIR)
for i, p in enumerate(paths):
# if i != 117:
# continue
print(i)
print(p)
_, file_name = os.path.split(p)
print(file_name)
content = read_all_content(p, encoding='gbk')
obj = json.loads(content)
save_image(obj['imageData'], i, file_name)
save_txt(obj['shapes'], i, file_name)
def delete_6():
paths = get_all_file_from_dir(PARENT_DIR_PATH)
for p in paths:
_, name = os.path.split(p)
label = name.split('-')[0]
is_move = False
for c in 'QWERTYUIOPLKJHGFDSAZXCVBNM':
if label.find(c) >= 0:
is_move = True
break
if is_move:
print(p)
shutil.move(p, DES_PATH + name)
def get_file_mapping(files_path, file_type='jpg'):
"""
get mapping
:param files_path:
:param file_type:
:return:
"""
mapping = {}
files_path = get_all_file_from_dir(files_path)
for p in files_path:
if p.find(file_type) >= 0:
_, name = os.path.split(p)
mapping[name.split('.')[0]] = p
return mapping
def main():
paths = get_all_file_from_dir(JSON_DIR)
for i, p in enumerate(paths):
# if i != 117:
# continue
if '.json' not in p:
continue
print(i)
print(p)
_, file_name = os.path.split(p)
file_name = file_name.replace('.json', '')
print(file_name)
content = read_all_content(p)
obj = json.loads(content)
save_image(obj['imageData'], i, file_name)
def move():
for dirpath, dirnames, filenames in os.walk(PARENT_DIR_PATH):
for d in dirnames:
try:
sub_dir = os.path.join(dirpath, d)
files = get_all_file_from_dir(sub_dir)
_, image_name = os.path.split(files[0])
try:
if len(files) <= 4:
for p in files:
shutil.move(p, DES_PATH + image_name)
print('delete' + sub_dir)
continue
except Exception as e:
print(e)
except Exception as e:
print(e)
def main():
paths = get_all_file_from_dir(JSON_DIR)
# try:
for i, p in enumerate(paths):
try:
if '.json' not in p:
continue
print(i)
if i < 0:
continue
content = read_all_content(p)
obj = json.loads(content)
file_name = os.path.split(p)[1].split('.')[0]
# file_name = get_file_name(obj['imageData'])
save_image(obj['imageData'], i, file_name)
save_txt(obj['shapes'], i, file_name)
except:
pass
def main():
txt_and_image_paths = 'D:/label_result_2020_3_10/label_result/箱门中文字识别/txt'
des_line_dir = 'D:/label_result_2020_3_10/label_result/箱门中文字识别/line/'
paths = get_all_file_from_dir(txt_and_image_paths)
try:
for i, p in enumerate(paths):
if '.jpg' not in p:
continue
img_name = os.path.split(p)[1].replace('.jpg', '')
img = read_image(p)
txt_path = p.replace('.jpg', '.txt')
points, is_error = get_points(txt_path)
if is_error:
print(p)
draw_image = get_detect_result(points, img)
write_image(des_line_dir + img_name + '.jpg', draw_image)
except Exception as e:
print(e)
def main():
for dirpath, dirnames, filenames in os.walk(num_path):
for d in dirnames:
sub_dir = os.path.join(dirpath, d)
files = get_all_file_from_dir(sub_dir)
if len(files) == 0:
# delete_file_in_dir(sub_dir)
print('delete' + sub_dir)
continue
_, image_name = os.path.split(files[0])
is_delete = False
for c in 'QWERTYUIOPLKJHGFDSAZXCVBNM':
if image_name.find(c) >= 0:
is_delete = True
break
if len(files) < 2:
for p in files:
_, file_name = os.path.split(p)
shutil.move(p, DES_path + file_name)
int(float(points[7]))]]
}
shapes.append(shape)
return shapes
def load_image_str(image_path):
with open(image_path, "rb") as imageFile:
image_s = base64.b64encode(imageFile.read())
image_s = str(image_s).replace('b\'', '')
image_s = str(image_s).replace('\'', '')
return str(image_s)
if __name__ == '__main__':
paths = get_all_file_from_dir(IMAGE_LABEL_DIR)
image_mapping = get_mapping(paths, file_type='jpg')
txt_mapping = get_mapping(paths, file_type='txt')
txt_mapping_keys = txt_mapping.keys()
index = 1
for key in image_mapping.keys():
if key in txt_mapping_keys:
try:
shapes = load_shapes(txt_mapping[key])
image_str = load_image_str(image_mapping[key])
json_obj = {
"flags": {},
"shapes": shapes,
"lineColor": [0, 255, 0, 128],
"fillColor": [255, 0, 0, 128],
"imagePath": "",
-------------------------------------------------
File Name： delete_empty_txt
Description :
Author : 'li'
date： 2018/10/14
-------------------------------------------------
Change Activity:
2018/10/14:
-------------------------------------------------
"""
import os
from utility.file_path_utility import get_all_file_from_dir
DIR_PATH = 'E:\dataset\detection/training_data/'
paths = get_all_file_from_dir(DIR_PATH)
for p in paths:
is_delete = False
if p.find('txt') > 0:
with open(p, encoding='utf8', mode='r') as file:
lines = file.readlines()
if len(lines) == 0:
is_delete = True
if is_delete:
_dir, name = os.path.split(p)
new_name = name.replace('txt', 'jpg')
try:
os.remove(p)
os.replace(DIR_PATH + name)
except Exception as e:
-------------------------------------------------
Change Activity:
2018/8/31:
-------------------------------------------------
"""
import json
import os
from utility.file_io_utility import read_all_content
from utility.file_path_utility import get_all_file_from_dir, create_dir
from xml.dom.minidom import Document
__author__ = 'li'
result_dir_path = 'C:/Users\lr\Desktop/123'
result_paths = get_all_file_from_dir(result_dir_path)
save_annotation_dir = './xml/'
create_dir(save_annotation_dir)
def load_result(result_paths):
"""
load result
:param result_paths:
:return:
"""
for p in result_paths:
if p.find('txt') > 0:
with open(p, mode='r', encoding='utf8') as file:
lines = file.readlines()
if len(lines) == 0:
# -*- coding: utf-8 -*-
"""
-------------------------------------------------
File Name： __init__.py
Description :
Author : 'li'
date： 2018/8/6
-------------------------------------------------
Change Activity:
2018/8/6:
-------------------------------------------------
"""
import os
import shutil
from utility.file_path_utility import get_all_file_from_dir
dir_path = 'E:\dataset\\11-25\horizontal/'
des_path = 'C:\\Users\lr\Desktop\small/'
paths = get_all_file_from_dir(dir_path)
for path in paths:
size = os.path.getsize(path)
if size < 700:
print(path)
_, file_name = os.path.split(path)
shutil.move(path, des_path + file_name)
# -*- coding: utf-8 -*-
"""
-------------------------------------------------
File Name： copy_hang_image
Description :
Author : 'li'
date： 2018/8/18
-------------------------------------------------
Change Activity:
2018/8/18:
-------------------------------------------------
"""
import shutil
import uuid
from utility.file_path_utility import get_all_file_from_dir
__author__ = 'li'
IMAGE_DIR = 'C:/Users\lr\Desktop\error/'
HANG_DIR = 'C:/Users\lr\Desktop\error_img/'
imgs = get_all_file_from_dir(IMAGE_DIR)
index = 1
for i in imgs:
index = index + 1
if i.find("txt") >= 0 or i.find("DMG") >= 0:
continue
new_path = HANG_DIR + str(uuid.uuid4()) + '.jpg'
shutil.copyfile(i, new_path)
-------------------------------------------------
Change Activity:
2018/8/1:
-------------------------------------------------
"""
import shutil
import cv2
from utility.file_path_utility import get_all_file_from_dir
__author__ = 'li'
"""
把不同形状的训练集做区分：水平的和垂直的。
"""
ORIGINAL_PATH = 'F:/dataset/container_dataset/text_area/'
HORIZONTAL_PATH = 'F:/dataset/container_dataset/text_area_horizontal/'
VERTICAL_PATH = 'F:/dataset/container_dataset/text_area_vertical/'
all_image_path = get_all_file_from_dir(ORIGINAL_PATH)
for path in all_image_path:
if path.find('.jpg') <= 0:
continue
img = cv2.imread(path)
shape = img.shape
image_name = path.split('\\')[-1]
if shape[0] >= shape[1]: # 高大于宽
shutil.copy(path, VERTICAL_PATH + image_name)
continue
shutil.copy(path, HORIZONTAL_PATH + image_name)
# -*- coding: utf-8 -*-
"""
-------------------------------------------------
File Name： data_validate
Description :
Author : 'li'
date： 2018/9/9
-------------------------------------------------
Change Activity:
2018/9/9:
-------------------------------------------------
"""
import os
from utility.file_path_utility import get_all_file_from_dir
__author__ = 'li'
images_path = 'F:\dataset\horizontal'
paths = get_all_file_from_dir(images_path)
for p in paths:
_, image_name = os.path.split(p)
label = image_name.split('-')[0]
for c in label:
if not ('1234567890'.find(c) >= 0 or
'QWERTYUIOPLKJHGFDSAZXCVBNM'.find(c) >= 0 or '#'.find(c) >= 0):
print(p)
File Name： validate_gen_txt
Description :
Author : 'li'
date： 2018/10/14
-------------------------------------------------
Change Activity:
2018/10/14:
-------------------------------------------------
"""
import numpy as np
from utility.file_path_utility import get_all_file_from_dir
TXT_DIR = 'E:\dataset\detection/training_data/'
paths = get_all_file_from_dir(TXT_DIR)
def polygon_area(poly):
'''
compute area of a polygon
:param poly:
:return:
'''
edge = [(poly[1][0] - poly[0][0]) * (poly[1][1] + poly[0][1]),
(poly[2][0] - poly[1][0]) * (poly[2][1] + poly[1][1]),
(poly[3][0] - poly[2][0]) * (poly[3][1] + poly[2][1]),
(poly[0][0] - poly[3][0]) * (poly[0][1] + poly[3][1])]
return np.sum(edge) / 2.
date： 2018/7/29
-------------------------------------------------
Change Activity:
2018/7/29:
-------------------------------------------------
"""
import shutil
from utility.file_path_utility import get_all_file_from_dir
__author__ = 'li'
img_path = 'F:/dataset/second_label/image/'
annotation_path = 'F:/dataset/second_label/detection_label/'
des_path = 'F:/dataset/second_label/des/'
img_path = get_all_file_from_dir(img_path)
annotation_path = get_all_file_from_dir(annotation_path)
def load_map(img_path):
map = {}
for path in img_path:
name = path.split('\\')[-1].split('.')[0]
map[name] = path
return map
img_map = load_map(img_path)
anno_map = load_map(annotation_path)
i = 1
for key in anno_map.keys():
-------------------------------------------------
File Name： six_image_move
Description :
Author : 'li'
date： 2018/9/23
-------------------------------------------------
Change Activity:
2018/9/23:
-------------------------------------------------
"""
import os
import shutil
from utility.delete_file import delete_file_in_dir
from utility.file_path_utility import get_all_file_from_dir
__author__ = 'li'
PARENT_DIR_PATH = 'E:\dataset/new_seg/6/'
DES_DIR_PATH = 'E:\dataset\six_uncheck/'
for dirpath, dirnames, filenames in os.walk(PARENT_DIR_PATH):
for d in dirnames:
sub_dir = os.path.join(dirpath, d)
files = get_all_file_from_dir(sub_dir)
if len(files) <= 2:
for f in files:
dir_p, name = os.path.split(f)
print(f)
shutil.move(f, DES_DIR_PATH + name)
