1. The test sets of COCO data set coco minival and coco test-dev:
The corresponding content of the two data sets on the official website is as follows:
COCO data set official website: https://cocodataset.org/#download
The reference website for distinguishing the two data sets: https://zhuanlan.zhihu.com/p/ 533676547
2. Commonly used conda commands:
①Create a virtual environment
conda create -n env-name python=xx -y
②Activate the virtual environment:
conda activate env-name
③Exit the virtual environment
conda deactivate
④Copy virtual environment
conda create -n 新环境名 --clone 旧环境名
⑤Export environment:
conda env export > env_name.yaml
⑥Use conda env export to export a file in yaml format. This file records the environment name, software source address and installation package list. In turn, use the yaml configuration file to create a new environment.
conda env create -f env_name.yaml
⑦Migrate virtual environment:
pip install conda-pack """安装conda-pack"""
conda pack -n env_name """env_name为环境名,打包之后的文件名为env_name.tar.gz"""
conda pack -n env_name --ignore-editable-packages """出现动态链接库无法打包时使用"""
Upload env_name.tar.gz to the anaconda3 environment of the new device, that is, in the conda/env/ directory
mkdir -p env_name """创建新环境名文件夹"""
tar -xzf env_name.tar.gz -C env_name """解压打包出来的压缩文件,压缩包解压到文件夹"""
conda activate env_name """"激活迁移的环境"""
3. The role of python -m
Reference: https://zhuanlan.zhihu.com/p/339104015
python xxx.py
python -m xxx.py
Here are two ways to load py files:
1 is called direct operation
2. Start the module as a script (note: but the value of __name__ is 'main')
python xxx.py -m xxxx
Experience with the second way:
It is equivalent to python will locate the path after -m in the code execution path, that is, the code in xxxx