string_util's Introduction

一个自己的库

dataHandle

保存数据

from string_util.dataHandle import save_data

save_data(路径，文件名前缀，保存的数据)
会自动检索路径中有该文件前缀的数据，并且每次调用创建一个后缀文件，比如说
citysacpe_0.pt
citysacpe_1.pt

读取数据

from string_util.dataHandle import dataset

dataset(路径，文件名前缀)
读取上面保存文件时保存的文件，符合pytorch dataset的定义

get_config

读取yml文件

from string_util import CONFIG
cfg = CONFIG("./haha.yml")
print(cfg.epoch_time)

显示内容

cfg.print()
cfg.haha.print()

结合argparse使用

from string_util import CONFIG
import argparse
cfg = CONFIG("./haha.yml")
parser = argparse.ArgumentParser(description='choose work')
parser.add_argument('-m', type=str,
                    default="train")
cfg.update(parser.parse_args().__dict__)
cfg.print()

time

获取程序运行时间

from string_util import TIME
import time

TIME.re_time(0, "训练开始", groud="train")
time.sleep(1)
TIME.re_time(1, "训练结束", groud="train")
TIME.show_time()

dataset

cityscapes

from string_util.dataset.cityscape import Cityscape_dataset
import torch

city_data_load = Cityscape_dataset("./datasets/cityscapes", split="val")
for i in range(city_data_load.__len__()):
    img ,lbl, path = city_data_load.__getitem__(i)

GTA

from string_util.dataset.gta import GTA_dataset
from tqdm import tqdm

gta_data_load = GTA_dataset("./datasets/GTA5", split="train")
for i in tqdm(range(gta_data_load.__len__())):
    img ,lbl,path = gta_data_load.__getitem__(i)

SYNTHIA

from string_util.dataset.synthia import Synthia_dataset
from tqdm import tqdm

synthia_data_load = Synthia_dataset("./datasets/RAND_CITYSCAPES", split="train", mean=[0,0,0])
for i in tqdm(range(synthia_data_load.__len__())):
    img ,lbl = synthia_data_load.__getitem__(i)

Recommend Projects