Frigate Yolov9-t quantization parameters

Hi, 
Using AMD huggingface yolox-s-int8 is working on frigate and I was to detect objects using a custom build from the source of onnxrunetime vitisai and migraphx. I need to try to use yolov9 on frigate but for some reason I failed to convert it to a16w8 to able to load most of the model on the NPU and not use the CPU at all. Would you please let me know to do it. Thanks 


