raykit is a python toolkit including a diverse set of useful functions for fast-prototyping in research, which is developed by Anyi Rao.

For fun only 🏖️

pip install raykit

raykit with version odd e.g. 0.0.3 is for simple usage, while version even e.g. 0.0.4 is for pytorch-based usage. you can specifiy it with e.g. pip install raykit==0.0.3

Requirements pytorch is needed for even number version.

from raykit.package import * Usually when we start a python program, we need to import a lot of packages such as argparse, os, os.path as osp, numpy as np etc. Just simply use from raykit.package import *, you get all.

When you are at the terminal, you want to check a file quickly but you don't have a GUI. Use the following. With read_txt_list, read_json, read_pkl, you read a txt as list or a json as dict or a pickle as it be. mmcv have the the similar functionalities, such as mmcv.load('test.json'), mmcv.load('test.yaml'), mmcv.load('test.pkl') but do not load txt.

With write_txt_list, write_json, write_pkl, you write as vice versa as above.

With get_folder_list, you os.listdir and output the list to a desired location.

With mkdir_ifmiss, you do not need to use os.makedirs(,exist_ok=True).

With strcal(string,num,fill) you are able to calculate a string type data and a float or int number and zfill it.

to_numpy and to_torch quickly transfer the datatype.

raykit changelog, mmcv