search

itsnine / yolov5-onnxruntime

YOLOv5 ONNX Runtime C++ inference code.
247 stars 58 forks source link

Half precision #7

Closed guishilike closed 2 years ago

guishilike commented 2 years ago

Official yolov5 PyTorch repo uses half precision. I try the onnx model with half precision on python, and speed increased. Can this repo support half precision？

itsnine commented 2 years ago

@guishilike hello, I haven't tried it yet, but if the half precision onnx model works in python onnxruntime inference, it should work in c++ version as well, if I'm not mistaken there are no additional flags for fp16 onnxruntime inference.