[pytorch environment installation]

pytorch install , Simple record

1. Get ready Anaconda Environmental Science , Download from the official website
conda After installation ,cmd It can be done directly conda Command operation

2. Environment configuration
（1） The environment name is paddle ,python Version is 3.9, Subsequent deep learning operations are carried out in this environment

conda create -n pytorch python=3.9

（2） Enter the designated environment

conda activate pytorch 

(3)pytorch Get the official website according to your own configuration conda Download instructions ,pytorch Download from the official website

For example, I downloaded cpu edition , It is recommended to use pip download , use conda because pytorch Servers are slow in foreign countries ,conda Download , hold -c The suffix of is deleted. It is faster to use the image source

pip3 install torch torchvision torchaudio

（4） After installation , Verify that the environment is installed successfully

import torch
a = torch.cuda.is_available()
print(a)

Activate the environment - Get into python Interpreter - load torch library , In fact, the installation was successful without any exception

3. download yolov5 Source code , stay github Direct download

pycharm Similarly, you can download it from the official website ,Pycharm , use pycharm Open the download paddledetection Engineering documents , Select corresponding python Interpreter File-Setting-Project-python interpreter choice conda Inside pytorch In the environment python.exe Interpreter .

After downloading, there are... In the directory requirements.txt, For installation python Other dependent Libraries , stay cmd perhaps pycharm Install under the console , My subsequent console operations are in pycharm perform , It's more convenient .

pip install -r requirements.txt

4. test pytorch-yolov5 The environment is set up

yolov5s.pt Is a model weight file , If not, you need to download

python detect.py --source data/images/bus.jpg --weights pretrained/yolov5s.pt