DmlExecutionProvider should be add

semihhdemirel / YOLOv11-ONNX-TensorRT

1 stars 0 forks source link

Open JuZi233 opened 1 month ago

JuZi233 commented 1 month ago

pip install onnxruntime-directml

then

self.session = ort.InferenceSession(model_byte.SerializeToString(), providers=["DmlExecutionProvider"],sess_options = options)

this could be faster fo : self.session.run..

and ,why postprocess cost so many time .

JuZi233 commented 1 month ago

i rewrite the preprocess , but postprocess