How do I setup and run the PyTorch yolov5 example with Anaconda?
- Install Anaconda
- Clone Juice-Labs/yolo from
git@github.com:Juice-Labs/yolo.git
All of the following steps are run from an Anaconda Prompt from within the directory that you cloned Juice-Labs/yolo to.
Create an Anaconda environment and install PyTorch and yolov5 dependencies by running the following from the cloned repository:
conda env create -n yolo --file environment.yml
conda activate yolo
pip install -r requirements.txt
Run yolov5 with native CUDA from the Anaconda Prompt, or with the Juice GPU if you've run the installer and the path containing python.exe is listed as active in the %LocalAppData%\Juice\juice.cfg:
python yolo.py
Run yolov5 through the Juice GPU from the Anaconda Prompt with the path to juicify.exe (included in JuiceClient.zip) in PATH:
juicify.exe %USERPROFILE%\anaconda3\envs\yolo\python.exe yolo.py