Open pinguvenom opened 1 year ago
I'm testing fastreid project in Here.
I'm using mgn_R50-ibn model referencing your config file.
converting pth-> onnx is succeed, but my testing similairty score result is entirely too high compared with your testing result.
I used sample code of this repository and sample images in samples/pedestrian_reid
Can you say reasons for this difference?
Thanks!
----- 230308 added ----- I make reasonable similiary score using your supported model.
But, I still don't know the difference between yours and my onnx model.
Can I take your advice?
I'm testing fastreid project in Here.
I'm using mgn_R50-ibn model referencing your config file.
converting pth-> onnx is succeed, but my testing similairty score result is entirely too high compared with your testing result.![Screenshot from 2023-03-08 10-51-12](https://user-images.githubusercontent.com/104476020/223598664-9ee7933d-af0a-4412-b8e7-3b726c7649aa.png)
I used sample code of this repository and sample images in samples/pedestrian_reid
Can you say reasons for this difference?
Thanks!
----- 230308 added ----- I make reasonable similiary score using your supported model.
But, I still don't know the difference between yours and my onnx model.
Can I take your advice?