difference between yolo3.weights and yolo3.onnx

[hoaquocphan]

Hi all,

I'm porting yolov3 to onnxruntime by code C.

I see that the Pre/Post processing steps in the guide of onnx/yolov3 same as your python code. (I think they copy code from here to make the guide for them) so the Pre/Post processing steps are the same.

1/ experiencor/keras-yolo3 use yolo3.weights -> data1 data1 + post processing step -> detected image. 2/ onnx/yolov3 use yolo3.onnx -> data2 data2 + post processing step -> detected image (error)

reason of error in (2) because of data1 and data2 are different.

so could anyone help me about this?