You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
124 lines
3.8 KiB
124 lines
3.8 KiB
""" |
|
验证图片尺寸和分离测试集（5%）和训练集（95%） |
|
初始化的时候使用，有新的图片后，可以把图片放在new目录里面使用。 |
|
""" |
|
|
|
import os |
|
import time |
|
import datetime |
|
import json |
|
import random |
|
import os.path |
|
import shutil |
|
|
|
import pymysql |
|
from PIL import Image |
|
|
|
|
|
def convertjpg(jpgfile, outdir, width=227, height=227): |
|
'''转换图片分辨率''' |
|
img=Image.open(jpgfile) |
|
try: |
|
new_img=img.resize((width,height),Image.BILINEAR) |
|
if img.mode == "P" or img.mode == "RGBA": |
|
new_img = new_img.convert('RGB') |
|
new_img.save(outdir) |
|
except Exception as e: |
|
print("图片转换失败",e) |
|
|
|
def spilt_train_test(origin_dir,train_dir,test_dir): |
|
'''将样本集分成9：1''' |
|
img_list = os.listdir(origin_dir) |
|
random.seed(time.time()) |
|
random.shuffle(img_list) |
|
R = int(len(img_list)*0.1) |
|
for file_name in img_list[:R]: |
|
src = os.path.join(origin_dir, file_name) |
|
dst = os.path.join(test_dir, file_name) |
|
shutil.move(src, dst) |
|
for file_name in img_list[R+1:]: |
|
src = os.path.join(origin_dir, file_name) |
|
dst = os.path.join(train_dir, file_name) |
|
shutil.move(src, dst) |
|
|
|
|
|
def get_date_list(start=None, end=None): |
|
'''获取两日期间日期列表''' |
|
data_list = [] |
|
datestart=datetime.datetime.strptime(start,'%Y-%m-%d') |
|
dateend=datetime.datetime.strptime(end,'%Y-%m-%d') |
|
while datestart<dateend: |
|
datestart+=datetime.timedelta(days=1) |
|
data_list.append(datestart.strftime('%Y-%m-%d')) |
|
return data_list |
|
|
|
|
|
def get_label(date,typeid): |
|
'''从获取数据库获取标签''' |
|
conn = pymysql.connect(host='ip', |
|
port=3306, |
|
user='**', |
|
password='**', |
|
db='**', |
|
charset='utf8') |
|
cursor = conn.cursor() |
|
# sql = "SELECT result,savedir FROM new_ocr_dir WHERE typeid = '{0}' AND time LIKE '{1}%'".format(typeid,date) |
|
sql = "SELECT result,savedir FROM new_ocr_dir WHERE typeid = {0} AND time LIKE '{1}%'".format(typeid, date) |
|
result = () |
|
try: |
|
cursor.execute(sql) |
|
result = cursor.fetchall() |
|
except Exception as e: |
|
print("查询数据库失败：{0}".format(e)) |
|
return result |
|
|
|
|
|
def set_label(label, dir, typid, id, date): |
|
''' |
|
设置标签并修改图片分辨率 |
|
:param label: 标签 |
|
:param dir: 图片原地址 |
|
:param outdir: 图片新地址 |
|
''' |
|
outdir = "data/{1}_{2}_{3}_{4}.jpg".format(typid, date, id, label) |
|
convertjpg(dir, outdir, 227, 227) |
|
# try: |
|
# with open(dir, 'rb') as f: |
|
# img = f.read() |
|
# with open("data/{0}/{1}_{2}_{3}_{4}.jpg".format(typid, typid, date, id, label), 'wb') as f: |
|
# f.write(img) |
|
# except Exception: |
|
# print(label, dir, typid) |
|
|
|
|
|
def solve_lable_dir(label_dir): |
|
'''提取标签和地址''' |
|
labels = [] |
|
dirs = [] |
|
for per in label_dir: |
|
label = json.loads(per[0]).get('result','') |
|
dir = per[1][13:] |
|
labels.append(label) |
|
dirs.append(dir) |
|
return labels, dirs |
|
|
|
|
|
def make_sample(): |
|
'''选定日期和类别，制作带标签的样本集''' |
|
dates = get_date_list('2019-10-12', '2019-12-08') |
|
typeids = ['3200','3060','3050','3040','3000','2050','2040','2000','1050','1040'] |
|
for typeid in typeids: |
|
for date in dates: |
|
print(date) |
|
label_dir = get_label(date, typeid) |
|
labels, dirs = solve_lable_dir(label_dir) |
|
for i in range(len(labels)): |
|
label = labels[i] |
|
label = label.replace('|','#') |
|
dir = dirs[i] |
|
set_label(label, dir, typeid, i, date) |
|
|
|
|
|
if __name__ == '__main__': |
|
make_sample() |
|
# spilt_train_test('data','train','test')
|
|
|