How to get more FPS when using PyTorch Hub.

[shershunov]

Without changing the number of parameters in the model.

[glenn-jocher]

@EkelviNistars 👋 Hello! Thanks for asking about improving inference speed. YOLOv5 🚀 can be run on CPU and GPU. When running inference on GPU the steps you can take to speed up inference are:

• Increase --batch-size for PyTorch Hub models
• Reduce --img-size
• Reduce model size, i.e. from YOLOv5x -> YOLOv5l -> YOLOv5m -> YOLOv5s
• Upgrade your hardware to a faster GPU

[shershunov]

How to set the batch-size parameter in PyTorch

[glenn-jocher]

results = model([im1, im2, im3, etc...])