How to use with custom YoloX model?

[Worse2Worst-Galileo]

Excuse me.

Issue Details

I tried to use this repo with my own YoloX model.
I have .onnx file of the model.
However, my model differ from the example model of this repo (YOLOX-Nano, 480x640).

1. My model's input is 640x640, while your model is 480x640.
2. My model output 7 classes, while the YOLOX-Nano outputs 80 classes.
   [Update]
   I tried training a new model to be the same size as YOLOX-Nano, 480x640.
   However, the model still cannot detect anything anyway.

(Please look into images for netron details of the models)
My Model

Your Model

I tried changing configs in "image_processor/detection_engine.cpp", but I can't make the bounding box appear on the screen.
I tried printing the "box_confidence" value, it's always very small.
So, I multiplied it by 100, the bounding boxes wouldn't appear anyway.
The program seems to work fine with YOLOX-Nano, 480x640 though.

Could you please tell me how to use the program with my own model?

Thank you very much.

[Worse2Worst-Galileo]

I found the root cause now.
It seems like my model takes image inputs as it is, not normalize.
I removed the normalize part, and it can detect now.